100% found this document useful (4 votes)
18 views

Learn Data Mining Through Excel: A Step-by-Step Approach for Understanding Machine Learning Methods, 2nd Edition Hong Zhou 2024 scribd download

Hong

Uploaded by

prietlvarm
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd
100% found this document useful (4 votes)
18 views

Learn Data Mining Through Excel: A Step-by-Step Approach for Understanding Machine Learning Methods, 2nd Edition Hong Zhou 2024 scribd download

Hong

Uploaded by

prietlvarm
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd
You are on page 1/ 66

Download Full Version ebookmass - Visit ebookmass.

com

Learn Data Mining Through Excel: A Step-by-Step


Approach for Understanding Machine Learning
Methods, 2nd Edition Hong Zhou

https://ebookmass.com/product/learn-data-mining-through-
excel-a-step-by-step-approach-for-understanding-machine-
learning-methods-2nd-edition-hong-zhou/

OR CLICK HERE

DOWLOAD NOW

Discover More Ebook - Explore Now at ebookmass.com


Instant digital products (PDF, ePub, MOBI) ready for you
Download now and discover formats that fit your needs...

Learn Programming with C: An Easy Step-by-Step Self-


Practice Book for Learning C 1st Edition Imran

https://ebookmass.com/product/learn-programming-with-c-an-easy-step-
by-step-self-practice-book-for-learning-c-1st-edition-imran/

ebookmass.com

Bayesian Statistics for Beginners: A Step-By-Step Approach


Therese M Donovan

https://ebookmass.com/product/bayesian-statistics-for-beginners-a-
step-by-step-approach-therese-m-donovan/

ebookmass.com

Elementary Statistics: A Step By Step Approach 10th


Edition, (Ebook PDF)

https://ebookmass.com/product/elementary-statistics-a-step-by-step-
approach-10th-edition-ebook-pdf/

ebookmass.com

Geisterschlucht (John Decker 7) (German Edition) Anthony


M. Strong

https://ebookmass.com/product/geisterschlucht-john-decker-7-german-
edition-anthony-m-strong/

ebookmass.com
Progress and Recent Trends in Microbial Fuel Cells 1st
Edition Patit Paban Kundu

https://ebookmass.com/product/progress-and-recent-trends-in-microbial-
fuel-cells-1st-edition-patit-paban-kundu/

ebookmass.com

Complete Solutions Manual for Calculus of a Single


Variable, Early Transcendental Functions 7th Edition Ron
Larson
https://ebookmass.com/product/complete-solutions-manual-for-calculus-
of-a-single-variable-early-transcendental-functions-7th-edition-ron-
larson/
ebookmass.com

False Start (Vegas Aces: The Playbook Book 2) Lisa Suzanne

https://ebookmass.com/product/false-start-vegas-aces-the-playbook-
book-2-lisa-suzanne/

ebookmass.com

Pediatric Cardiology, An Issue of Pediatric Clinics of


North America (Volume 67-5) (The Clinics: Internal
Medicine, Volume 67-5) Ivy Dunbar
https://ebookmass.com/product/pediatric-cardiology-an-issue-of-
pediatric-clinics-of-north-america-volume-67-5-the-clinics-internal-
medicine-volume-67-5-ivy-dunbar/
ebookmass.com

Standard of Living: Essays on Economics, History, and


Religion in Honor of John E. Murray Patrick Gray

https://ebookmass.com/product/standard-of-living-essays-on-economics-
history-and-religion-in-honor-of-john-e-murray-patrick-gray/

ebookmass.com
Danger Close: Domestic Extremist Threat #1 Comes Clean
Patrick Byrne

https://ebookmass.com/product/danger-close-domestic-extremist-
threat-1-comes-clean-patrick-byrne/

ebookmass.com
Learn Data Mining
Through Excel
A Step-by-Step Approach for
Understanding Machine
Learning Methods
Second Edition

Hong Zhou
Learn Data Mining Through Excel: A Step-by-Step Approach for Understanding
Machine Learning Methods
Hong Zhou
Department of Mathematics and Computer Science
University of Saint Joseph, West Hartford, CT, USA

ISBN-13 (pbk): 978-1-4842-9770-4 ISBN-13 (electronic): 978-1-4842-9771-1


https://doi.org/10.1007/978-1-4842-9771-1

Copyright © 2023 by Hong Zhou


This work is subject to copyright. All rights are reserved by the Publisher, whether the whole or part of the
material is concerned, specifically the rights of translation, reprinting, reuse of illustrations, recitation,
broadcasting, reproduction on microfilms or in any other physical way, and transmission or information
storage and retrieval, electronic adaptation, computer software, or by similar or dissimilar methodology now
known or hereafter developed.
Trademarked names, logos, and images may appear in this book. Rather than use a trademark symbol with
every occurrence of a trademarked name, logo, or image we use the names, logos, and images only in an
editorial fashion and to the benefit of the trademark owner, with no intention of infringement of the
trademark.
The use in this publication of trade names, trademarks, service marks, and similar terms, even if they are not
identified as such, is not to be taken as an expression of opinion as to whether or not they are subject to
proprietary rights.
While the advice and information in this book are believed to be true and accurate at the date of publication,
neither the authors nor the editors nor the publisher can accept any legal responsibility for any errors or
omissions that may be made. The publisher makes no warranty, express or implied, with respect to the
material contained herein.
Managing Director, Apress Media LLC: Welmoed Spahr
Acquisitions Editor: Mark Powers
Development Editor: Laura Berendson
Editorial Project Manager: Shaul Elson
Cover designed by eStudioCalamar
Cover image by Vecstock on Freepik (www.freepik.com)
Distributed to the book trade worldwide by Springer Science+Business Media New York, 1 New York Plaza,
Suite 4600, New York, NY 10004-1562, USA. Phone 1-800-SPRINGER, fax (201) 348-4505, e-mail orders-ny@
springer-sbm.com, or visit www.springeronline.com. Apress Media, LLC is a California LLC and the sole
member (owner) is Springer Science + Business Media Finance Inc (SSBM Finance Inc). SSBM Finance Inc
is a Delaware corporation.
For information on translations, please e-mail booktranslations@springernature.com; for reprint,
paperback, or audio rights, please e-mail bookpermissions@springernature.com.
Apress titles may be purchased in bulk for academic, corporate, or promotional use. eBook versions and
licenses are also available for most titles. For more information, reference our Print and eBook Bulk Sales
web page at http://www.apress.com/bulk-sales.
Any source code or other supplementary material referenced by the author in this book is available to
readers on GitHub (github.com/apress). For more detailed information, please visit https://www.apress.
com/gp/services/source-code.
Paper in this product is recyclable.
For my family and friends.
Table of Contents
About the Author����������������������������������������������������������������������������������������������������� ix

About the Technical Reviewer��������������������������������������������������������������������������������� xi

Chapter 1: Excel and Data Mining���������������������������������������������������������������������������� 1


Why Excel?������������������������������������������������������������������������������������������������������������������������������������ 1
Why the Second Edition?�������������������������������������������������������������������������������������������������������������� 5
Prepare Some Excel Skills������������������������������������������������������������������������������������������������������������ 7
Formula����������������������������������������������������������������������������������������������������������������������������������� 7
Autofill or Copy������������������������������������������������������������������������������������������������������������������������ 8
Absolute Reference����������������������������������������������������������������������������������������������������������������� 9
Paste Special and Paste Values��������������������������������������������������������������������������������������������� 12
IF Function Series������������������������������������������������������������������������������������������������������������������ 14
Reinforcement Exercises������������������������������������������������������������������������������������������������������������ 21
Review Points����������������������������������������������������������������������������������������������������������������������������� 21

Chapter 2: Linear Regression��������������������������������������������������������������������������������� 23


General Understanding��������������������������������������������������������������������������������������������������������������� 23
Learn Linear Regression Through Excel�������������������������������������������������������������������������������������� 27
Learn Multiple Linear Regression Through Excel������������������������������������������������������������������������ 30
Reinforcement Exercises������������������������������������������������������������������������������������������������������������ 34
Review Points����������������������������������������������������������������������������������������������������������������������������� 35

Chapter 3: K-Means Clustering������������������������������������������������������������������������������� 37


General Understanding��������������������������������������������������������������������������������������������������������������� 37
Learn K-Means Clustering Through Excel����������������������������������������������������������������������������������� 38
Reinforcement Exercises������������������������������������������������������������������������������������������������������������ 51
Review Points����������������������������������������������������������������������������������������������������������������������������� 52

v
Table of Contents

Chapter 4: Linear Discriminant Analysis���������������������������������������������������������������� 53


General Understanding��������������������������������������������������������������������������������������������������������������� 53
Solver������������������������������������������������������������������������������������������������������������������������������������������ 55
Learn LDA Through Excel������������������������������������������������������������������������������������������������������������ 59
Reinforcement Exercises������������������������������������������������������������������������������������������������������������ 70
Review Points����������������������������������������������������������������������������������������������������������������������������� 70

Chapter 5: Cross-Validation and ROC��������������������������������������������������������������������� 71


General Understanding of Cross-Validation�������������������������������������������������������������������������������� 71
Learn Cross-Validation Through Excel���������������������������������������������������������������������������������������� 72
General Understanding of ROC Analysis������������������������������������������������������������������������������������� 77
Learn ROC Analysis Through Excel��������������������������������������������������������������������������������������������� 79
Reinforcement Exercises������������������������������������������������������������������������������������������������������������ 84
Review Points����������������������������������������������������������������������������������������������������������������������������� 85

Chapter 6: Logistic Regression������������������������������������������������������������������������������� 87


General Understanding��������������������������������������������������������������������������������������������������������������� 87
Learn Logistic Regression Through Excel����������������������������������������������������������������������������������� 89
By Means of Maximizing Log Likelihoods������������������������������������������������������������������������������ 90
By Means of Minimizing Log Losses������������������������������������������������������������������������������������� 97
Reinforcement Exercises���������������������������������������������������������������������������������������������������������� 100
Review Points��������������������������������������������������������������������������������������������������������������������������� 101

Chapter 7: K-Nearest Neighbors��������������������������������������������������������������������������� 103


General Understanding������������������������������������������������������������������������������������������������������������� 103
Learn K-NN Through Excel�������������������������������������������������������������������������������������������������������� 104
Experiment 1����������������������������������������������������������������������������������������������������������������������� 105
Experiment 2����������������������������������������������������������������������������������������������������������������������� 108
Experiment 3����������������������������������������������������������������������������������������������������������������������� 113
Experiment 4����������������������������������������������������������������������������������������������������������������������� 118
Reinforcement Exercises���������������������������������������������������������������������������������������������������������� 120
Review Points��������������������������������������������������������������������������������������������������������������������������� 120

vi
Table of Contents

Chapter 8: Hierarchical Clustering and Dendrogram�������������������������������������������� 121


General Understanding������������������������������������������������������������������������������������������������������������� 121
Draw a Dendrogram in Excel Without Add-ins�������������������������������������������������������������������������� 123
Learn Hierarchical Clustering Through Excel���������������������������������������������������������������������������� 129
Reinforcement Exercises���������������������������������������������������������������������������������������������������������� 141
Review Points��������������������������������������������������������������������������������������������������������������������������� 142

Chapter 9: Naive Bayes Classification������������������������������������������������������������������ 143


General Understanding������������������������������������������������������������������������������������������������������������� 143
Learn Naïve Bayes Through Excel��������������������������������������������������������������������������������������������� 148
Exercise 1���������������������������������������������������������������������������������������������������������������������������� 148
Exercise 2���������������������������������������������������������������������������������������������������������������������������� 152
Reinforcement Exercises���������������������������������������������������������������������������������������������������������� 159
Review Points��������������������������������������������������������������������������������������������������������������������������� 159

Chapter 10: Decision Trees����������������������������������������������������������������������������������� 161


General Understanding������������������������������������������������������������������������������������������������������������� 162
Learn Decision Trees Through Excel����������������������������������������������������������������������������������������� 165
Learn Decision Trees Through Excel������������������������������������������������������������������������������������ 166
A Better Approach���������������������������������������������������������������������������������������������������������������� 178
Apply the Model������������������������������������������������������������������������������������������������������������������� 182
Reinforcement Exercises���������������������������������������������������������������������������������������������������������� 185
Review Points��������������������������������������������������������������������������������������������������������������������������� 185

Chapter 11: EDA, Data Cleaning, and Feature Selection��������������������������������������� 187


Learn Exploratory Data Analysis Through Excel������������������������������������������������������������������������ 187
Learn Data Cleaning Through Excel������������������������������������������������������������������������������������������ 196
Feature Selection���������������������������������������������������������������������������������������������������������������������� 200
Normalization and Standardization������������������������������������������������������������������������������������� 201
Correlation��������������������������������������������������������������������������������������������������������������������������� 204
Learn Feature Selection Through Excel������������������������������������������������������������������������������� 205
Reinforcement Exercises���������������������������������������������������������������������������������������������������������� 209
Review Points��������������������������������������������������������������������������������������������������������������������������� 209
vii
Table of Contents

Chapter 12: Association Analysis������������������������������������������������������������������������� 211


General Understanding������������������������������������������������������������������������������������������������������������� 211
Learn Association Analysis Through Excel�������������������������������������������������������������������������������� 214
Reinforcement Exercises���������������������������������������������������������������������������������������������������������� 222
Review Points��������������������������������������������������������������������������������������������������������������������������� 223

Chapter 13: Artificial Neural Network������������������������������������������������������������������ 225


General Understanding������������������������������������������������������������������������������������������������������������� 225
Learn Neural Network Through Excel��������������������������������������������������������������������������������������� 227
Experiment 1����������������������������������������������������������������������������������������������������������������������� 227
Experiment 2����������������������������������������������������������������������������������������������������������������������� 239
Reinforcement Exercises���������������������������������������������������������������������������������������������������������� 251
Review Points��������������������������������������������������������������������������������������������������������������������������� 251

Chapter 14: Text Mining���������������������������������������������������������������������������������������� 253


General Understanding������������������������������������������������������������������������������������������������������������� 253
Learn Text Mining Through Excel���������������������������������������������������������������������������������������������� 255
Word Embedding and ChatGPT������������������������������������������������������������������������������������������������� 272
Reinforcement Exercises���������������������������������������������������������������������������������������������������������� 274
Review Points��������������������������������������������������������������������������������������������������������������������������� 274

Chapter 15: After Excel����������������������������������������������������������������������������������������� 277

Index��������������������������������������������������������������������������������������������������������������������� 281

viii
About the Author
Hong Zhou, PhD, is a professor of computer science and
mathematics and has been teaching courses in computer
science, data science, mathematics, and informatics at the
University of Saint Joseph for nearly 20 years. His research
interests include bioinformatics, data mining, software
agents, and blockchain. Prior to his current position, he was
a Java developer in Silicon Valley. Dr. Zhou believes that
learners can develop a better foundation of data mining
models when they visually experience them step by step,
which is what Excel offers. He has employed Excel in teaching data mining and finds it
an effective approach for both data mining learners and educators.

ix
About the Technical Reviewer
Adam Gladstone has over 25 years’ experience in software
development, mostly in C++ and C#. He has worked mainly
in the investment banking and finance sectors. For the last
few years, he has been developing data science and machine
learning skills, particularly in Python and R after completing
a degree in maths and statistics. He loves programming in
C++ and C# and his free time is spent developing software
tools.

xi
CHAPTER 1

Excel and Data Mining


Let’s get right to the topic. Why do we need to learn Excel in our data mining endeavor?
It is true that there are some outstanding data mining software tools such as RapidMiner
that make the mining process easy and straightforward. In addition, programming
languages Python and R have a large number of reliable packages dedicated to various
data mining tasks. What is the purpose of studying data mining or machine learning
through Excel?

Why Excel?
If you are already an experienced data mining professional, I would say that you
are asking the right question and probably you should not read this book. However,
if you are a beginner in data mining, or a visual learner, or want to understand the
mathematical background behind some popular data mining techniques, or an
educator, then this book is right for you, and probably is the first book you should read
before you start your data mining journey.
Excel allows you to work with data in a transparent manner, meaning when an Excel
file is opened, the data is visible immediately and every step of data processing is also
visible. Intermediate results are contained in the Excel worksheet and can be examined
while you are conducting your mining task. This allows you to obtain a deep and clear
understanding of how the data are manipulated and how the results are obtained.
Other software tools and programming languages hide critical aspects of the model
construction process. For most data mining projects, the goal is to find the internal
hidden patterns inside the data. Therefore, hiding the detailed process is beneficial to the
users of the tools or packages. But it is not helpful for beginners, visual learners, or those
who want to understand how the mining process works. Let me use k-nearest neighbors
method (K-NN) to illustrate the learning differences between RapidMiner, R, and Excel.
Before we do that, we need to understand several terminologies in data mining.

1
© Hong Zhou 2023
H. Zhou, Learn Data Mining Through Excel, https://doi.org/10.1007/978-1-4842-9771-1_1
Chapter 1 Excel and Data Mining

There are two types of data mining techniques: supervised and unsupervised.
Supervised methods require the use of a training dataset to “train” the software
programs or algorithms (such programs or algorithms are often referred to as machines)
first. Programs are trained to reach an optimal state called a model. This is why a
training process is also called modeling. Data mining methods can also be categorized
into parametric and nonparametric methods. For parametric methods, a model is just
a set of parameters or rules obtained through the training process that are believed to
allow the programs to work well with the training dataset. Nonparametric methods do
not generate a set of parameters. Instead, they dynamically evaluate the incoming data
based on the existing dataset. You may be confused by such definitions at this time. They
will make sense soon.
What is a training dataset? In a training dataset, the target variable (also called label,
target, dependent variable, outcome variable, response), the value of which is to be
predicted, is given or known. The value of the target variable depends on the values of
other variables which are usually called attributes, predictors, or independent variables.
Based on the attribute values, a supervised data mining method computes (or so-called
predicts) the value of the target variable. Some computed target values might not match
the known target values in the training dataset. A good model indicates an optimal set of
parameters or rules that can minimize the mismatches.
A model is usually constructed to work on future datasets with unknown target
values in a supervised data mining method. Such future datasets are commonly called
scoring datasets. In an unsupervised data mining method, however, there is no training
dataset and the model is an algorithm that can directly be applied on the scoring
datasets. K-nearest neighbors method is a supervised data mining technique.
Suppose we want to predict if a person is likely to accept a credit card offer based
on the person’s age, gender, income, and number of credit cards they already have. The
target variable is the response to the credit card offer (assume it is either Yes or No),
while age, gender, income, and number of existing credit cards are the attributes. In the
training dataset, all variables including both the target and attributes are known. In such
a scenario, a K-NN model is constructed through the use of the training dataset. Based
on the constructed model, we can predict the responses to the credit card offer of people
whose information is stored in the scoring dataset.

2
Chapter 1 Excel and Data Mining

In RapidMiner, one of the best data mining tools, the prediction process is as follows:
retrieve both the training data and scoring data from the repository ➤ set role for the
training data ➤ apply the K-NN operator on the training data to construct the model ➤
connect the model and the scoring data to the Apply Model operator. That’s it! You can
now execute the process and the result is obtained. Yes, very straightforward. This is
shown in Figure 1-1. Be aware that there is no model validation in this simple process.

Figure 1-1. K-NN model in RapidMiner

Applying K-NN method is very simple in R, too. After loading the library “class”, read
the training data and scoring data, make use of the K-NN function, and by then we have
finished our job: ready to view our result. This is demonstrated in Figure 1-2. Note that
lines starting with “#” are comments.

3
Chapter 1 Excel and Data Mining

Figure 1-2. K-NN in R

The knowledge you have gained from the preceding tasks is enough to just be able
to apply the data mining method K-NN. But if you are trying to understand, step by step,
why and how K-NN works, you will need a lot more information. Excel can offer you the
opportunity to go through a step-by-step analysis process on a dataset during which you
can develop a solid understanding of the K-NN algorithm. With this solid understanding,
you can then be more proficient in using other powerful tools or programming
languages. Most importantly, you will have a better understanding of the quality and
value of your data mining results. You will see that in later chapters.
Of course, Excel is much more limited in data mining compared to Python, R, and
RapidMiner. Excel can only work with data up to a smaller size limit. Meanwhile, some
data mining techniques are too complicated to be practiced through Excel. Nonetheless,
Excel provides us direct and visual understanding of the data mining mechanisms. In
addition, Excel is naturally suitable for data preparation.

4
Chapter 1 Excel and Data Mining

Today, because of the software tools and other packages, most effort in a data mining
task is spent on understanding the task (including the business understanding and data
understanding), preparing the data, and presenting the results. Less than 10% of the
effort is spent on the modeling process. The process of preparing the data for modeling is
called data engineering. Excel has an advantage on data engineering when the datasets
are not too large because it can give us a visual representation of data engineering, which
allows us to be more confident in our data preparation process.
As an experienced educator, I realize that students can better develop a deep
understanding of data mining methods if these methods are also explained through
step-by-step instructions in Excel. Studying through Excel unveils the mystery behind
data mining or machine learning methods and makes students more confident in
applying these methods.
Did I just mention machine learning? Yes, I did. Machine learning is another buzz
phrase today. What is machine learning? What is the difference between data mining
and machine learning? Moreover, what is artificial intelligence (AI) and what is the
difference between AI and machine learning?
The purposes of machine learning and data mining are somewhat different. The
purpose of machine learning is to study how computers can develop human-like
learning ability by learning from data. The purpose of data mining is to find valuable
patterns or knowledge from data. However, data mining makes use of machine learning
methods to achieve its goals, that is, the methodologies of data mining and machine
learning can be the same. Anyhow, it is not necessary to differentiate them, and I would
suggest that we treat them the same at this moment.
The same scenario applies to AI and machine learning. Simply put, AI is computer
software that simulates human brain functions, while machine learning trains computer
algorithm through data to mimic human thinking ability. Because most AI makes use of
machine learning to achieve its goals, machine learning is usually considered a subset of
AI and therefore it is almost impossible to differentiate AI and machine learning either.

Why the Second Edition?


Practice is the key in learning, especially in learning technologies including Excel and
data mining. Thus, the first reason for the second edition is to provide homework-like
exercise(s) at the end of each chapter so that you can practice on your own the core skills

5
Chapter 1 Excel and Data Mining

explained in each chapter. Starting from this first chapter, you will find reinforcement
exercises which may even be more challenging than whatever was explained step by step
in the book. Of course standard answers or suggestions are provided to these exercises.
The second reason is that some chapters are updated based on feedbacks from some
readers.

1. Chapter 3, K-Means Clustering: By introducing the function


SHEET, the k-means clustering process is updated so that the
process is even more automatic.

2. Chapter 6, Logistic Regression: As loss function is frequently


used in machine learning, the concept and use of loss function is
introduced in this chapter.

3. Chapter 9, Naive Bayes Classification: The Naïve Bayes’ theorem


is now explained in detail such that the theorem is deduced
from the conditional probability. This gives readers a clear
mathematical understanding about why and how the Naïve Bayes
classification works.

4. Chapter 14, Text Mining: Because of the arrival of ChatGPT, I


consider it is necessary to give a simple and quick introduction
about the word vectorization (also called text embedding)
technology which is widely used in large language models
including ChatGPT.

The third reason is that some minor errors are fixed and some formulas are updated.
For example, the function IFERROR is used more often to simplify a few formulas.
The last reason, probably the primary reason for the second edition, is that there are
two new chapters.

1. One chapter is about exploratory data analysis (EDA), data


cleaning, and feature selection. These critical concepts and skills
are a must for studying data mining or machine learning. They are
usually introduced before any data mining or machine learning
models are explained. Though I take this opportunity to include
them in this edition, I introduce them in Chapter 11 after several
machine learning models. The reason is explained in Chapter 11.
So, you will figure out why when you read Chapter 11.

6
Chapter 1 Excel and Data Mining

2. Another new chapter (Chapter 8) is about hierarchical clustering


which was once mentioned together with k-means clustering
in the first edition of this book. This chapter presents two bright
highlights. One is to implement a bottom-up hierarchical
clustering algorithm in Excel; the second is to show how to draw a
dendrogram in Excel without any add-in.

Prepare Some Excel Skills


There are quite some Excel skills to learn in this book. I will explain some of them in
detail when we need to use them. However, there are several fundamental Excel skills
and functions that we need to be familiar with before we start talking about data mining.

Formula
Formula is the most important feature of Excel. Writing a formula is like writing a
programming statement. In Excel, a formula always starts with an equal sign (“=”
without quotation marks).
Upon opening an Excel file, we are greeted with a table-like worksheet. Yes, every
worksheet is a huge table. One reason why Excel is naturally suitable for data storage,
analysis, and mining is because data are automatically arranged in a table format in
Excel. Each cell in the big table has a name or so-called reference. By default, each
column is labeled by a letter, while each row is labeled with a number. For example, the
very first cell at the top-left corner is cell A1, that is, column A and row 1. The content in a
cell, whatever it is, is represented by the cell reference.
Enter number 1 in cell A1. The value of cell A1 is 1 and A1 represents 1 at
this moment.
Enter the formula "=A1*10" (without the double quotation marks and no space
before "=") in cell B1 and hit the Enter key. Note that the formula starts with “=”.
Be aware that this is the only time a formula is presented inside a pair of double
quotation marks in this book. From now on, all formulas are presented directly without
quotation marks.
Enter the text “A1 * 10” in cell C1. Because the text does not start with “=”, it is not a
formula.

7
Chapter 1 Excel and Data Mining

Our worksheet looks like Figure 1-3.

Figure 1-3. Excel formula

Autofill or Copy
Autofill is another critical feature of Excel which makes Excel capable of working with a
relatively large dataset. Autofill is also called “copy” by many people.
Let’s learn autofill by the following experiment:

1. Enter 1 in cell A1.

2. Enter 2 in cell A2.

3. Select both cells A1 and A2.

4. Release the left mouse button.

5. Move the mouse cursor to the lower right corner of cell A2 until
the cursor becomes a black cross (shown in Figure 1-4).

Figure 1-4. Cursor becomes a black cross

6. Press down the left mouse button and drag down to cell A6.

The cells A1:A6 are automatically filled with numbers 1, 2, 3, 4, 5, and 6. This process
is called autofill. Some people call it “copy”, too. But more precisely, this process is
autofill.

8
Chapter 1 Excel and Data Mining

Let’s conduct another experiment:

1. Select cell B1 (make sure that B1 still has the formula =A1*10). Lift
up the left mouse button.

2. Move the mouse cursor to the lower right corner of cell B1 until
the cursor becomes a black cross.

3. Drag down the mouse cursor to cell B6. Our worksheet looks like Figure 1-5.

Figure 1-5. Autofill formula (relative reference)

Note that in Figure 1-5, the formula in cell B1 is =A1*10, but the formula in cell B2 is
automatically changed to =A2*10. When we autofill vertically down, Excel automatically
increments the row index of all the cell references by 1 in a formula. Similarly, when we
autofill horizontally to the right, Excel automatically increments the column index of all
the cell references in a formula.

Click on cell B3; the formula inside must be =A3*10.

Note Changing the number in A1 to 10 automatically changes the number in


B1 to 100. By default, the whole Excel workbook will automatically recalculate if a
cell value is changed. This is another important feature of Excel.

Absolute Reference
Assume that there is a ledger keeping the fixed interest rate and the amounts of loans
lent to customers. Our job is to compute the interest each customer owes. Follow these
instructions to complete the experiment:
9
Chapter 1 Excel and Data Mining

1. Open a blank Excel worksheet. Enter texts “Loan”, “Interest”, and


“Rate” in cells A1, B1, and D1. Leave C1 blank. Enter “5%” inside
cell D2 (without quotation marks).

2. In cell A2, enter the formula =RANDBETWEEN(1000,50000). This


formula randomly generates an integer between 1000 and 50000
inclusively.

3. Autofill from A2 to A12.

4. In cell B2, enter the formula =A2*D2. This formula calculates how
much interest A2 owes. Everything should be perfectly fine at
this moment.

5. Let’s try another quick autofill skill: select cell B2 ➤ move the
mouse cursor to the lower right corner of cell B2 until the cursor
becomes a black cross ➤ double-click. See Figure 1-6. This
double-click action automatically fills in formulas from cell B3 to
cell B12 (B2 already has a formula). Note: autofill by double-click
works only with vertical downward autofill operations.

Figure 1-6. Autofill by double-click

10
Chapter 1 Excel and Data Mining

Our worksheet now looks like Figure 1-7.

Figure 1-7. Autofill failed on interest calculation

Except for cell B2, all other cells in column B do not obtain the correct result. The
reason is because when we autofill from B2 to B12 vertically, the original formula in cell
B2 (=A2*D2) changes: the row index of every cell reference in the formula is automatically
incremented one by one. As shown in Figure 1-7, the formula in cell B3 is =A3*D3. We can
imagine that the formula in cell B4 is =A4*D4.
In cell B3, we need the formula to be =A3*D2, that is, when we autofill from B2 to B3,
we need the cell reference A2 to be changed to A3, but we need the cell reference D2 to
stay the same.
As mentioned before, cell reference D2 has two parts: the column index D and the
row index 2. For vertical autofill, the column index will never change but the row index
will. To keep the row index unchanged during autofill, we need to place “$” before the
row index. This means, the formula in cell B2 should be =A2*D$2. Using the symbol “$”
to keep cell reference(s) in a formula unchanged or locked in autofill operations is called
absolute reference.
With this corrected formula in cell B2, let’s autofill from cell B2 to B12 again. We
should get the correct result this time.

11
Chapter 1 Excel and Data Mining

Be aware that in cell B2, the formula =A2*$D$2 works fine, too, for this specific task.
When placing $ before the column index D, the column index won’t change even when
we are performing a horizontal autofill. However, be aware that we will run into cases
where we must keep only part of a cell reference (either row index or column index only)
unchanged in an autofill operation.

Paste Special and Paste Values


Continue with the worksheet on which we have been trying some interest calculation.
Because they are generated by the function RANDBETWEEN, the numbers in column A
keep changing in our worksheet and this is not good for demonstration. We would like to
copy the numbers in cells A1:D12 to another location and remove the random number
feature. We can accomplish this task by using the Paste Special feature.
Follow these instructions:

1. Select the cell area A1:D12. Under the main tab Home, click tool
tab Copy. Let’s copy the numbers to area A15:D26.

2. Click on cell A15; under the main tab Home, click Paste ➤ select
Paste Special. This process is explained in Figure 1-8. Note if your
operating system has Right Click enabled, you can right-click on
cell A15 ➤ select Paste Special, too.

12
Chapter 1 Excel and Data Mining

Figure 1-8. Select Paste Special

3. A small menu shows up. On this menu, choose Values under


Paste as shown in Figure 1-9. Be aware of another available
feature “Transpose” on this menu. It is a good idea to practice this
Transpose feature as it is very useful for data preparation.

Figure 1-9. Paste Values

13
Chapter 1 Excel and Data Mining

Part of our worksheet looks similar to Figure 1-10.

Figure 1-10. After Paste Values

IF Function Series


The IF statement is said to be the most used statement in programming, and this saying
is also true in our learning data mining through Excel effort. As we are going to make use
of the function IF and other IF-related functions very often, it is a good idea for us to get
some basic understanding of them first.
This book comes with a number of sample Excel files. We are going to make use
of them often. These files are available at https://github.com/hhohho/Learn-Data-
Mining-through-Excel-2. These Excel files are examples designed for different data
mining practices. There are two ways to download them:

1. Download them together as a compressed file and decompress


them into a folder (directory) on your computer.

2. Download them based on individual chapters.

After downloading, open the file Chapter1-1a.xlsx. Our worksheet looks like
Figure 1-11. If you want, you can enter the data manually in a blank worksheet, too.

14
Chapter 1 Excel and Data Mining

Figure 1-11. The performances of some sales representatives

Our first task is to compute the commissions for each sales representative. The
commission is calculated in this way:

1. Any sale amount lower or equal to $50000 has no commission.

2. The sale amount more than $50000 but less than or equal to
$100000 gains 10% commission.

3. The sale amount more than $100000 gains 20% commission.

Our formula to compute the commissions is based on the IF function. Enter the
following formula in cell D2:

=IF(C2<=50000,0,IF(C2<=100000,(C2-50000)*10%,50000*0.1+(C2-100000)*20%))

IF function has the basic form: IF(boolean-expression, true-value, false-value). In


our formula, if the sales in C2 are no more than 50000, 0 is returned. However, when C2
is more than 50000, a nested IF function is used here as we have two more conditions
to assess.
In the inner IF function, if C2<=100000, the amount more than 50000 is multiplied
by 0.1 (10%) and returned. Otherwise, C2 must be more than 100000, and therefore the
calculated result of the expression 50000*0.1+(C2-100000)*20%) is returned.
Our worksheet looks exactly like Figure 1-12.

15
Chapter 1 Excel and Data Mining

Figure 1-12. Commissions calculated by using the IF function

What if we want to count how many sales representatives have sales between
50000 and 100000 in season 1? To answer this question, we need to apply the function
COUNTIFS. In Excel, there is another function named COUNTIF. COUNTIFS is capable
of every task that COUNTIF is capable of, and there are many tasks for which we must
use COUNTIFS instead of COUNTIF.
The explanation of COUNTIFS from Microsoft Office Support states “The COUNTIFS
function applies criteria to cells across multiple ranges and counts the number of times
all criteria are met.” Its syntax is like COUNTIFS(criteria_range1, criteria_1, criteria_
range2, criteria_2, …). Based on this understanding, enter the following formula in
cell E1:

=COUNTIFS(B2:B10,1,C2:C10,">50000",C2:C10,"<=100000")

This formula makes sure that

• In cell range B2:B10, the season is 1.

• And in cell range C2:C10, the sales are >50000. Note the criteria is
placed inside a pair of double quotation marks.

• And in cell range C2:C10, the sales are <=100000.


The answer in cell E1 must be 2.
We want to examine the total sales and probably the averaged sales for every sales
representative. For this purpose, we need to set up a table. Follow these instructions to
set up the table:

16
Chapter 1 Excel and Data Mining

1. Copy cells A1:A10 to F1:F10. Select cells F1:F10; follow


instructions in Figure 1-13 to select Remove Duplicates (click
main tab Data ➤ click the icon for Remove Duplicates).

Figure 1-13. Select the Remove Duplicates feature

2. A small menu may show up like Figure 1-14. Make sure to select
“Continue with the current selection”. Click the button Remove
Duplicates.

Figure 1-14. Continue with the current selection

17
Chapter 1 Excel and Data Mining

3. Check “My data has headers” as shown in Figure 1-15.

Figure 1-15. My data has headers

4. Click OK on the upcoming small menu.

5. Enter the text “Yearly Sales” in cell G1 and “Average Sales” in


cell H1.

Our worksheet should look like Figure 1-16.

Figure 1-16. Table setup

18
Chapter 1 Excel and Data Mining

For yearly sales computation, we need to use the function SUMIFS. Again, there is
another function named SUMIF. The syntaxes of these two functions are different and
could cause confusion. As SUMIFS can accomplish anything SUMIF can do, let’s stick to
SUMIFS for simplicity.

1. SUMIFS has the syntax SUMIFS(sum_range, criteria_range1,


criteria1, criteria_range2, criteria2, …). Based on this
understanding, enter the following formula in cell G2:

=SUMIFS(C$2:C$10,A$2:A$10,F2)

The preceding formula works this way: in the array A2:A10, cells
A2, A7, A9, and A10 equal F2 (“Amie”); thus, their corresponding
cells in the array C2:C10 are summed, that is, cells C2, C7, C9, and
C10 are summed. Observe the use of absolute cell references in
the formula because we are going to autofill from cell G2 to G5.

2. Autofill from cell G2 to cell G5.

3. Enter the following formula in cell H2 (here you can use either
AVERAGEIF or AVERAGEIFS, they have similar syntax):

=AVERAGEIFS(C$2:C$10,A$2:A$10,F2)

4. Autofill from cell H2 to cell H5.

Our worksheet looks exactly like Figure 1-17. The complete result is available in
Chapter1-1b.xlsx.

19
Chapter 1 Excel and Data Mining

Figure 1-17. The use of functions SUMIFS and AVERAGEIFS

Beware that since Office 2019 or Microsoft 365, Excel adds two more functions into
the IF-function family: MAXIFS and MINIFS. Their syntaxes are very much similar to
those of COUNTIFS, SUMIFS, and AVERAGEIFS.
Another two Excel features used often in data preparation are data import and
export. To export data in Excel is fairly simple. For example, click File ➤ Save As ➤
select CSV to export the current worksheet into a file in CSV format. Note, CSV format
is probably the most used file format for data science. Here, CSV represents “comma-­
separated values.”
Today, Excel is capable of importing data from different resources. Click the main
tab Data ➤ click Get Data; a list of data sources is presented including From File,
From Databases, etc. For general purpose, importing data from files might be the most
common. Be aware that Excel can only import text files properly. It cannot properly
import files with special formatting such as Microsoft Word files.
Different Excel versions have different user interfaces on data importing. I won’t
illustrate how to use this feature in this book because all needed data for this book are
well prepared in those Excel files which can be downloaded from GitHub at
https://github.com/hhohho/Learn-Data-Mining-through-Excel-2.
Before I end this chapter, I would like to point out that this book is for Office 2016
and after. Thus, some new and more powerful functions of Office 365 are not introduced
in this book because they are not available in Office 2016.

20
Chapter 1 Excel and Data Mining

Reinforcement Exercises
If you would like to practice some of the skills mentioned in this chapter, please open the
Chapter1-HW.xlsx file to practice. Should you run into difficulties, please take a look at
the solutions inside Chapter1-HW-withAnswers.xlsx.

Review Points
This is our first chapter. Please review the following topics and skills:

1. The advantages of learning data mining through Excel examples

2. Excel formula

3. Excel skill autofill

4. Absolute and relative references

5. Remove Duplicates and Paste Special

6. Excel functions RANDBETWEEN, IF, COUNTIFS, SUMIFS, and


AVERAGEIFS

21
CHAPTER 2

Linear Regression
Please download the sample Excel files from https://github.com/hhohho/Learn-Data-
Mining-through-Excel-2 for this chapter’s exercises.

General Understanding
Linear regression is a predictive model in which training data is employed to construct
a linear model to make predictions on the scoring data. When we talk about a linear
model, we mean that the relationship between the target (dependent variable) and the
attribute(s) (independent variables) is linear. It is a convention to use the terminologies
“independent variable” and “dependent variable” in regression analysis. Therefore,
in this chapter, we will replace attribute with independent variable and substitute
dependent variable for target.
There might be one or more independent variables in linear regression analysis.
When there is only one independent variable, the linear model is expressed by the
commonly seen linear function y = mx + b, where y is the dependent variable, m is
the slope of the line, and b is the y-intercept. In most cases, there is more than one
independent variable; therefore, the linear model is represented as y = m1x1 + m2x2 + ...
+ mnxn + b, where there are n independent variables and mi is the coefficient associated
with a specific independent variable xi, for i = 1 ... n. To construct such a linear
model, we need to find the values of mi and b, based on the known y and xi values in the
training dataset.
Let’s start a scenario to learn what linear regression can do. In a southern beach
town, Tommy, the manager of a supermarket is thinking about predicting ice cream sales
based on the weather forecast. He has collected some data that relate weekly averaged
daily high temperatures to ice cream sales during a summer season. The training data
are presented in Table 2-1.

23
© Hong Zhou 2023
H. Zhou, Learn Data Mining Through Excel, https://doi.org/10.1007/978-1-4842-9771-1_2
Chapter 2 Linear Regression

Table 2-1. Ice cream sale vs. temperature


Temperature (F) Ice Cream Sale (Thousands
of Dollars)

91 89.8
87 90.2
86 81.1
88 83.0
92.8 90.9
95.2 119.0
93.3 94.9
97.7 132.4

Enter the preceding data into an Excel worksheet. Observe that in this scenario, the
Temperature is the only independent variable, and the Ice Cream Sale is the dependent
variable. Based on the given data, Tommy draws a scatter plot. The chart looks like
Figure 2-1.

Ice Cream Sale (dollar in thousand)


140
120
100
80
Sales

60
40
20
0
84 86 88 90 92 94 96 98 100
Temperatures (F)

Figure 2-1. Temperatures vs. ice cream sales

Based on the chart in Figure 2-1, if the average of the daily high temperatures of next
week is predicted to be 88.8 Fahrenheit degrees, Tommy finds it difficult to predict the
sales since there is no dot directly matching to the temperature 88.8. Right-click on a dot;
a small menu should show up. On the menu, click Add Trendline as shown in Figure 2-2.

24
Chapter 2 Linear Regression

Figure 2-2. Add Trendline

A trendline shows up in the chart as illustrated by Figure 2-3. Be aware that by


default, the line in Figure 2-3 is not a solid red line and the equation won’t show up.
It would be a good exercise to figure out how to format the trendline to be a solid red
line and how to display the equation on the chart. To display the equation on the chart,
right-­click on the trendline in the chart ➤ select Format Trendline… ➤ check “Display
Equation on chart”.

Ice Cream Sale (dollar in thousand)


140
120 y = 3.8394x - 253.17
100
80
Sales

60
40
20
0
84 86 88 90 92 94 96 98 100
Temperatures (F)

Figure 2-3. Using trendline to predict

With this linear line, Tommy can then estimate that when the temperature = 88.8,
the sales would be about $89k. Using the equation y = 3.8394x − 253.17, where m = 3.8394
and b = -253.17, Tommy can predict his ice cream sales more precisely:

3.8394 × 88.8 – 253.17 = $87.8k.


25
Chapter 2 Linear Regression

The linear equation here is the linear data mining model for this specific linear
regression study. We understand that for this type of linear equation y = mx + b,
m and b are the two determinants. Once m and b are found, the model is constructed and
finalized. Certainly, for this simple case, the model is represented by m and b, that is, the
parameter set composed of m and b is the model. The model construction process is to
find m and b.
How are m and b obtained? From Figure 2-3, we can tell that the trendline does not
pass through every point, it only follows the trend. There could be multiple distinct
trendlines if several people are drawing it manually and independently. How does Excel
come up with this specific trendline?
To find a specific linear equation in the form of y = mx + b, the least square method
is employed. Let me explain what the least square method is.
Observe the point corresponding to temperature = 91, the dot is not on the trendline.
This indicates that there is an error (we can understand it as difference) between
the trendline (the predicted value) and the actual data. In fact, all the data points in
Figure 2-3 have small errors. The sum of the squares of the errors can be represented as
n
E = ∑ ( yi − ( mxi + b ) )
2
(2-1)
i =1 

where yi is the actual ice cream sale corresponding to the temperature value xi in the
training dataset. The goal is to minimize the sum of the errors. For this purpose, we can
take the partial derivatives:

∂E ∂E (2-2)
= 0 and =0
∂x ∂y 

By solving the preceding two partial derivative equations, we can obtain the values of
m and b as

m=
n
( ∑ x )( ∑ y )
n∑ i =1xi yi −
n
i =1 i
n
i =1 i
(2-3)
n (∑ x ) − (∑ x )
n n 2
2
i =1 i  i =1 i

26
Chapter 2 Linear Regression

b=
( ∑ x )( ∑ y ) − ( ∑ x )( ∑
n
i =1 i
2 n
i =1 i
n
i =1 i
n
x yi
i =1 i ) (2-4)
n (∑ x ) − (∑ x )
n n 2
2
i =1 i i =1 i 

Learn Linear Regression Through Excel


In Equations (2-3) and (2-4), xi is the known averaged daily high temperature and yi is
the actual ice cream sale corresponding to xi. They are from the training dataset shown
in Table 2-1.
Open the file Chapter2-1a.xlsx. Enter the texts and formulas in columns C and D
as shown in Figure 2-4 to compute the values of m and b. The formulas in Figure 2-4 are
based on Equations (2-3) and (2-4). Observe that in Figure 2-4, temperature is renamed
as x and ice cream sale is renamed as y so that the setup follows Equations (2-3) and
(2-4) more closely.

Figure 2-4. Computing m and b via least square approximation

In Figure 2-4, A:A and B:B are used to represent all cells in column A and column
B, respectively. Thus, =SUM(A:A) and =SUM(B:B) will sum up all cells in column A and
column B, respectively. Also, an Excel function SUMPRODUCT is used to simplify the
computation between two arrays. An important understanding of SUMPRODUCT is that
the two arrays must be of the same type and length. Here, the same type means that the
two arrays can be both a column and a row. With both m and b computed, Tommy can
accomplish his predictions as shown in Figure 2-5 (enter the formula =D$8*C12 + D$9
inside cell D12 and autofill from D12 to D14).

27
Chapter 2 Linear Regression

Figure 2-5. Prediction based on a linear regression model

A much more efficient way to compute m and b is to make use of the two functions
SLOPE and INTERCEPT for this specific example. Follow these instructions to practice
the use of the two functions:

1. Input the x and y values as shown Figure 2-6. Note: input x and y
values only. Columns C and D are empty.

2. Enter m in cell C2 and b in cell D2.

3. Enter the formula =SLOPE(B2:B9, A2:A9) in cell C3.

28
Chapter 2 Linear Regression

Figure 2-6. Prepare data to use functions SLOPE and INTERCEPT

4. Enter the formula =INTERCEPT(B2:B9, A2:A9) in cell D3.

The result should look like Figure 2-7.

Figure 2-7. The result obtained by the SLOPE and INTERCEPT functions

Another way to obtain the values of m and b is through the Data Analysis tool of Excel.
This is a popular approach to take by many people, though I dislike it. The disadvantage
of using Data Analysis tool for regression is that once the coefficients and intercept value
are obtained, they won’t update along with the changes in the training data.
By default, our Excel does not have Data Analysis tool available. To make it available,
we must enable the built-in add-in Analysis ToolPak in our Excel. I will show you how to
enable Analysis ToolPak and Solver in our Excel later when we need to use Solver. As of
right now, we do not need to use the Data Analysis tool.

29
Chapter 2 Linear Regression

The results of the preceding learning processes are stored in the file
Chapter2-1b.xlsx.

Learn Multiple Linear Regression Through Excel


Tommy has successfully predicted his ice cream sales, but he wants to improve his
prediction even more. After some studies, he collected additional information such as
the number of tourists (obtained from hotels) and the number of sunny days in a week.
His updated training dataset is presented in Table 2-2.

Table 2-2. Ice cream sale vs. temperature, tourists, and sunny days
Temperature (F) Tourists Sunny Days Ice Cream Sale (dollar in thousand)

91 998 4 89.8
87 1256 7 90.2
86 791 6 81.1
88 705 5 83
92.8 1089 3 90.9
95.2 1135 6 119
93.3 1076 4 94.9
97.7 1198 7 132.4

There are three independent variables in Table 2-2: temperature, tourists, and
sunny days. When there is more than one independent variable, the linear regression
is specifically called multiple linear regression. Let’s denote temperature as x1, tourists
as x2, sunny days as x3, the y-intercept as b, and the sales as y. Our multiple linear
regression equation is y = m1x1 + m2x2 + m3x3 + b. For such multiple linear regression, we
need to make use of LINEST function to obtain the coefficients mi and b. Follow these
instructions to practice multiple linear regression through Excel:

1. Enter the data of Table 2-2 into cells A1:D9 in a new Excel
worksheet, or open the file Chapter2-2a.xlsx.

30
Chapter 2 Linear Regression

The model construction process is to find the optimal values of m1, m2,
m3, and b. We are going to make use of the array function LINEST. Be
advised that LINEST returns an array of values. The returned values
are in the order of mn, mn-1, …, m1, and b, the reversed order to the way in
which the formula is expressed. This necessitates the use of the INDEX
function to fetch individual values from the returned array. The INDEX
function is designed to work with a matrix or table. It requires a row
index and a column index to locate an element in a table. As LINEST
returns an array which has only one row, the row index is always 1,
while the column indexes are 1, 2, 3, and 4, respectively.

2. Enter the texts “Sunny Days”, “Tourists”, “Temperature”, and


“Y-intercept” in cells A11:A14.

3. Enter numbers 1, 2, 3, and 4 in cells B11:B14. Our worksheet


looks like Figure 2-8. I will explain why we need to enter these
numbers soon.

Figure 2-8. Set up the table to find coefficients and y-intercept

31
Random documents with unrelated
content Scribd suggests to you:
Abb. 3. Relief aus dem Grab des Ma-nofer. Berlin.
Die meisten späteren Epochen malten Bilder mit starker
Tiefenwirkung, mit Darstellung weiter Landschaften und Räume an
die Wand, in die sie gleichsam hineinführten; so nahmen sie ihr ihren
Sinn als Raumbegrenzung, dem Raum selbst seine
Geschlossenheit. Der Ägypter begnügt sich damit, die Darstellungen
von Göttermythen, von Königstaten, vom Besitz des Verstorbenen
als wagerecht fortlaufende Friesreihen in flachem, bemaltem Relief
zu geben. Denn er denkt logisch von der Wandfläche aus, deren
Wert nicht gemindert werden darf, deren Hauptbegrenzungen durch
Erdboden und Decke die wagerechte, ruhige Aufteilung verlangen,
deren fortlaufende Fläche die ununterbrochene Reihung fordert.
Diese Betonung der Wagerechten gibt dem Raum die Ruhe,
während die vertikalen Linien, die von der Erde aufstreben, Träger
architektonischer Bewegung sind. Es ist kein Zufall, daß gerade in
der untersten Friesreihe, von der Abbildung 3 nur den Anfang geben
kann, die Doppelreihe des Kleingeflügels fast wie ein Saum die
Wand abschließt, mit Tieren, deren Kleinheit und Zartheit den
schweren Schritt der Rinder über ihnen um so wuchtiger erscheinen
läßt. Die Fläche schreibt auch der Einzelform das Gesetz vor.
Nirgends treten die Darstellungen aus ihr heraus. Sie gibt dem
Ägypter die Oberfläche seiner Figuren (Abb. 3). Um ihre Konturen
herum tieft er den Grund aus, doch so, daß dieser Grund wieder
eine vollkommene Ebene bildet (Relief en creux). Er projiziert also
seine Gestalten vollkommen in die Fläche und nimmt ihnen jede
Tiefenwirkung. Sie schreiten im Profil, das am leichtesten die
Bewegung auf die Linie zurückzuführen erlaubt, gehen auf einem
Strich, statt auf einem Fußboden, und der Ort, an dem der Vorgang
sich abspielt, wird durch keinen Hintergrund angedeutet. Und doch
besitzen diese Künstler ein äußerst feines Naturgefühl, das aber
völlig im Stilbewußtsein aufgeht. Man sehe, wie auf einem Relief des
alten Reiches aus dem Grabe des Ma-nofer, wenn dem
Verstorbenen sein Besitz vorgeführt wird (Abb. 3), das schwere
Schreiten des Stieres, die elegante Zeichnung der Stelzvögel, der
steife Gang der Enten gegeben sind. Man verfolge die Halslinien der
prachtvollen Kranichgruppe vom Kopf bis herab zur Brust, den
Kontur des Stieres in Brust und Rücken. Die Modellierung der
Tierleiber innerhalb der Fläche des Reliefs ist von so vollkommener
Feinheit, daß man diese ganz leisen Hebungen und Senkungen,
diese feinsten Bewegungen der Fläche fast mit der Hand abtasten
muß, wenn man sie ganz genießen will. Sie sind absichtlich so zart
modelliert, und auch die Bemalung gab einst nur Farben und keine
Eigenschaften der Haut. Denn für das Auge muß der Kontur das
wichtigste bleiben, der als scharfe Linie in Licht und Schatten sich
am Rande der Gestalten zeichnet und das Relief der Zeichnung
gleichstellt. Die Linie ist die beherrschende Ausdrucksform. Sie
erklärt auch jenes sonderbare Vorschieben einer Schulter in der
ägyptischen Profildarstellung des Menschen. Es ist allerdings
fraglos, daß bei Darstellungen der ältesten Zeit die Tendenz dahin
ging, die Figur möglichst in allen Teilen klarzustellen, so daß man
den Kopf im Profil, die Brust von vorn und die Füße wieder im Profil
zeichnete. Aber es handelt sich dabei nicht um die Klarheit der
Körperflächen, sondern vielmehr der Umrisse, des Kontur. Es will
scheinen, als sei der Hauptgrund für die Vorschiebung der Achsel,
etwa bei dem Schreiber in Abb. 3, die Absicht gewesen, die beiden
Arme in der Zeichnung klar voneinander zu lösen. Die Beine gaben
diese Klarheit durch die Schreitstellung von selbst her, die Stellung
der Arme zueinander aber konnte so nicht ausgedrückt werden, da
sich zwischen sie der Körper schob. Der Ägypter sah also
zeichnerisch, wenn er als Maß der Entfernung vor allem die
Achsellinie nahm und die Entfernung vom Halsansatz durch sie
ausdrückte. Dem Bauch fehlt dagegen die Klarheit. Man merkt
seinen Linien, die nur Übergänge zur Profilstellung der Beine sind,
die Verlegenheit um logischen Ausdruck allzu deutlich an, und hier
versagt auch die Modellierung des Körpers. Dafür, daß bei dieser
Zeichnung nur Arm- und Schulterlinien maßgebend sind, nicht die
Klarstellung der Flächen, scheint es ein vollkommener Beweis zu
sein, daß, wenn beim dahinschreitenden Stier die Hörner
symmetrisch in der Fläche dargestellt werden, wie auf unserem
Relief, nur der obere Rand der Stirn mit ihnen in Vorderansicht
gestellt wird, ohne daß der Kopf dieser Richtung folgt.
Abb. 4. Tempel von Luksor. Fassade.
Abb. 5. Ammontempel zu Karnak. Schnitt durch die Halle (Papyrussäulen mit
geöffneten und geschlossenen Dolden).
Abb. 6. König Seti I. räuchernd. Relief des neuen
Reiches.
Über diese Epoche des alten Reiches hinaus schreitet die
Entwicklung weiter. Wie wir in ihr den Entwicklungsgang am
Grabbau verfolgen konnten, so zeigt von nun an der Tempelbau den
Weg. Schon im mittleren Reich, das um das Jahr 2000 zu setzen ist,
war die Technik des Pyramidenbaues immer schlechter geworden,
da der König nicht mehr absolut herrschte, nicht mehr so
unumschränkt über das ganze Volk gebot. Das neue Reich, das das
2. Jahrtausend v. Chr. in seiner zweiten Hälfte ausfüllt, kennt die
Pyramiden überhaupt nicht mehr und stellt den Tempel in den
Mittelpunkt seines architektonischen Interesses. Aber es erstrebt
nicht mehr Formklarheit, sondern seelische Wirkungen, bestattet
deshalb die Fürsten in düstere Felsengrüfte, deren Wände mit
geheimnisvollen Götterszenen und Todesmythen gefüllt werden, und
formt den Tempel als große ungegliederte Masse, um die Seele des
Andächtigen zu erschüttern. Die Formen sind allerdings, wie stets,
aus den vorhergehenden entwickelt, und man kann sagen, daß
diese Richtung auf maßstäbliche Größe eine Folge der
vorhergehenden, auf innere Größe gerichteten war. Die drei
Haupträume des Tempels, den Säulenhof, den großen Saal und den
Kultraum dahinter, hatte schon der Totentempel des alten Reiches
besessen; aber die Zeiten differenzieren sich sofort, wenn man die
feinfühlige Gliederung eines Torbaues im alten Reich mit der
Fassade eines Tempels, wie etwa des von Luksor, vergleicht, der der
Zeit kurz vor dem Beginn des neuen Reiches angehören dürfte. An
Stelle der zweckvollen Gliederungen ist die gesammelte Energie des
Ausdruckes getreten. Darauf beruht die ungeheure Kraft dieser
Bauwerke, daß ihre Art so gewaltsam ist gegenüber der Natur des
Bodens, in dem sie wurzeln. In Luksor, einem typischen Beispiel
(Abb. 4), ist die Fassade rein als Masse empfunden; die Wand, mit
flächenhaften, teppichartig gedachten Darstellungsreihen
übersponnen, die aus dem Leben der Könige und den Mythen der
Götter entnommen werden, spricht als große Fläche, und die
Begrenzung muß um so schärfer sein, muß die Wand energisch
zusammenschließen, da allein der Kontur dem ganzen Gebilde den
Halt gibt. So ist es empfunden, wenn die Begrenzungslinien von
unten aufsteigen, nach oben zu einander sich nähern und die Wand
zwischen sich immer enger zusammendrängen, bis die wagerechte
Hohlkehle oder eigentlich das von ihr weit vorgeschobene Sims den
Bau oben gegen den Luftraum abschließt. Dekorativ empfindende
Zeiten haben immer versucht, durch Vasen, Statuen und andere
Schmuckformen, die sie auf das obere Sims stellten, eine
allmähliche Überführung in den Luftraum herzustellen, und betonten
das Hauptportal dadurch, daß sie über ihm das Dach am höchsten
hinaufführten. Der Ägypter verfährt gerade umgekehrt. Er schließt
den Bau in strenger Horizontale und betont das Portal dadurch, daß
er die Wand hier niedriger abschließt. Dieses Herausheben soll nur
dem Nahenden den Eingang weisen. Die ruhende Mauerfläche muß
trotz seiner das Beherrschende bleiben. So nimmt man dem Portal
möglichst seinen Wert als gliedernde Unterbrechung der
Mauerfläche und führt die Einsenkung durch schräg aufsteigende
Linien den seitlichen Pylonen wieder zu. Genau ebenso wertet die
ägyptische Architektur im Innenbau des Tempels die Tür, die aus
dem Säulensaal zum Allerheiligsten führte, und die nur wenigen
Auserwählten sich öffnete. Und doch mußten in einer Halle, wie etwa
der des Ammontempels zu Karnak (Abb. 5), diese kolossalen Säulen
des Mittelganges, die wie riesige Wächter vor der geheimnisvollen
Tür standen, das Auge des Beschauers durch ihre Reihung gerade
dorthin führen, wie sie noch heute unseren Blick unwillkürlich dorthin
leiten. Es war der religiöse Zweck der ägyptischen Kunst, auf das
Geheimnisvolle hinzuweisen, ohne es zu entschleiern. Aber in der
Architektur der Halle, in den gewaltigen Massen des Säulensaales
kommt diese Tür kaum zum Wort gegenüber dem Gesamteindruck.
Denn neben diesen führenden Säulenreihen ziehen andere dahin,
und das Auge trifft nirgends einen Raum, der nicht nach wenigen
Schritten begrenzt wäre. Ein solcher Raum ist gar nicht als Halle in
unserem Sinn aufzufassen. Er ist eine gewaltige Masse von Säulen,
zwischen denen ein schmaler Raum zum Wandeln bleibt. Nicht der
freie Raum war die Hauptsache, sondern die atemraubende
Dichtigkeit der sperrenden Architektur. Es ist also das Prinzip des
möglichst wuchtigen Eindrucks im Innenbau dasselbe wie im
Außenbau. Aber es ist selbstverständlich, daß auch beim Innenraum
das Prinzip sich allmählich ausgebildet hat. Vergleicht man diese
Halle der 19. Dynastie (um 1300 v. Chr.) mit Hallen aus früherer Zeit,
so haben diese viel mehr Luftraum, viel mehr Bewegungsfreiheit für
den Wandelnden. Dabei bedingt die Stilentwicklung, daß auch die
Säule selbst früher empfindlichere Formen hatte. Wenn die Säulen
des alten Reiches (Abb. 1) in Schaft und Kapitell nach Maßgabe der
zugrunde liegenden Pflanzen durchgegliedert waren, so ist es für die
Säulen des Tempels in Karnak ganz unwichtig, daß auch ihnen
Pflanzenformen zugrunde liegen, die geöffnete Papyrusdolde den
Säulen des Mittelganges, die geschlossene denen der begleitenden
Reihen. Die Säule ist ein Steinklotz geworden, und die
Pflanzenmotive sind nur noch traditionell beibehalten. Die Säulen
sind zu ungeheurer Zahl gehäuft, aber sie scheinen als Masse viel
zu tragfähig im Verhältnis zur wirklichen Last. Unausgenützte Kraft
aber wirkt als Trägheit. Eine Masse ist noch kein organisches
Gebilde. Der tektonisch Denkende verlangt den klaren Ausdruck der
Verhältnisse zwischen Kraft und Last im Raume sowohl wie in der
einzelnen Säule.
Es ist notwendige Stileinheit, daß wie die Architektur so auch
Plastik und Malerei im späteren Reich allmählich zu summarischer
Behandlung der Formen kommen (Abb. 6). Man hat nicht einmal ein
Recht, von Erstarrung zu sprechen. Die Entwicklung zum
Gewaltigen verlangt das Vernachlässigen des Details. Der große
Eindruck der spätägyptischen Wand würde gelähmt sein, wenn ihr
Bilderschmuck nahe Betrachtung forderte. Die Plastik soll den
Beschauer erdrücken wie die Baukunst. Auch ihr Streben geht nach
der Monumentalität des maßstäblich Großen; Kolossalstatuen und
Riesenreliefs entstehen wie nie zuvor, Vergrößerungen der alten
durch die Religion geheiligten und geforderten Typen. Aber ihre
Formen sind verarmt, weil kein schöpferisches Gefühl mehr
stilbildend tätig ist, sind zur Phrase geworden, weil der Künstler
keine Naturform mehr erlebt, um sie zur Kunstform umzumünzen.
Daher die Unfähigkeit, diese Gestalten mit Formleben zu erfüllen,
ihre Leere, ihr Glattwerden oder das unwillkürliche, verstohlene
Eindringen realistischer Alltagszüge, eines Lächelns oder Drohens,
das deutlich zeigt, wie wenig man noch die Augenstrenge hat, die
die klassische Kunst des alten Reiches schuf. Daß dann der
rücksichtslose Naturalismus, in der Malerei sogar der
Impressionismus, der stärkste Gegensatz zur altägyptischen
Formstrenge, von den Ägäischen Inseln eindringend, zum
Durchbruch kommt, sobald ihm König Amenophis’ IV. Echnatons
Kampf gegen die alte Götterreligion für kurze Zeit Lebensmöglichkeit
schafft, ist nur natürliche Konsequenz. Aber es handelt sich hier um
eine reine Episode. Eine wirkliche Stilentwicklung findet späterhin
nicht mehr statt; nur daß die herben, hieratisch strengen Formen
immer weichlicher, immer ausdrucksloser werden, je mehr die
gefühlvolle Liebe unschöpferischer Perioden sie festzuhalten strebt.
Im wesentlichen finden noch die fremden Völker, die seit dem 6.
Jahrhundert v. Chr. Ägypten abwechselnd unterjochen, finden
Alexander der Große und seine Nachfolger bis zu den römischen
Kaisern in Ägypten dieselbe Kunst vor wie der Perser Kambyses, als
er im Jahre 525 v. Chr. Ägyptens letzten Pharao entthronte. Die
Usurpatoren fügten sich der Tradition des Landes, bauten ihre
Tempel dort im alten Stil und erschienen in der Plastik als Nachfolger
der alten Pharaonen in Kostüm, Haltung und Handlung; und so
bewahrte die offizielle ägyptische Kunst, deren große Taten dem
alten und mittleren Reich gehörten, ein langes Scheinleben, indes
längst neue Völker mit neuer Kraft eine Welt der Schönheit in Leben
und Kunst geschaffen hatten.
Zweites Kapitel.
Die prähistorische Kunst des
Ägäischen Meeres.
Es ist kein Zufall, daß die ägyptische Kunst, sobald ihre
Entwicklung sie für die impressionistische Auffassung reif gemacht
hatte, unter den Einfluß der sog. mykenischen Kultur geriet. Sie ist
die glänzendste Erscheinung der Bronzezeit. Der augenfrische
Impressionismus des Urmenschen hat hier, auf später hellenischem
Boden, Werke von überraschender Ausdrucksenergie geschaffen. In
Malerei und Goldschmiedekunst, in Elfenbeinplastik und Steinschnitt
gibt sie Tiere und Menschen in den kompliziertesten Bewegungen,
schafft die zartesten, feinlinigsten Ornamente und belebt mit ihnen
die bewegteste Architektur und das graziöseste Kunstgewerbe. Es
ist jene Kultur, von deren Schönheit die Epen Homers erzählen. Von
Kreta aus, dem Ort ihres Ursprungs und ihrem Hauptsitz, sandte sie
ihren Reichtum in die Welt. Diese strömende Gestaltungskraft, die
aus der Fülle ihres Reichtums Gestalten und Werke schuf, welche
noch heute mit der Unmittelbarkeit des künstlerisch Geschauten zu
uns sprechen, hat damals die ganze Welt befruchtet bis nach
Spanien hin und ins vorisraelitische Kanaan. Und in den
griechischen Mythen spielt König Minos von Kreta eine Rolle, die
diese Macht auch die Nachwelt ahnen ließ.
Abb. 7. Palast von Knossos. Treppenaufgang.
Abb. 8. Jüngling ein Gefäß tragend.
Fresko im Palaste von Knossos.
Ein stärkerer Gegensatz als zwischen ägyptischen und
kretischen Bauten (Abb. 7) ist kaum denkbar. Der Ägypter schließt
jeden Raum streng gegen den anderen ab, überall soll man sich
umgrenzt, eingehegt fühlen; der kretische Palast von Knossos öffnet
die Räume gegeneinander, um durch die formenreiche Folge
malerischer Durchblicke das Auge anzuregen. Die ägyptische Säule
ist wuchtige, oft überschwere Trägerin; die kretische, unten
zugespitzt, scheint zu balancieren, und ihr wulstiges Kapitell wirkt
wie ein Pfühl, der sich unter dem Druck elastisch biegt. Die
ägyptische Säule, die das Tragen stärker ausdrückt, ist also
struktiver geformt als die kretische. Und ebenso ist die ägyptische
Wand strenge Begrenzung des Zimmers, während die kretische es
liebt, sich mit räumlich tiefen Darstellungen zu schmücken. Es
genügt, eine der kretischen Wandmalereien aus demselben Palast
von Knossos (Abb. 8) mit einem ägyptischen Relief der frühen, von
fremdem Einfluß unberührten Zeit (Abb. 3) zu vergleichen, um die
Gegensätzlichkeit zweier Kunstanschauungen zu empfinden. Der
Ägypter gräbt die Linien des Umrisses, wie er sie bei genauer
Detailbetrachtung feststellt, in die Fläche und sucht durch flache
Modellierung der Innenform Herr zu werden. Der Kreter gibt den
Gesamteindruck, die „Impression“, gibt die Form, wie sie dem Auge
beim ersten Blick erscheint, und vernachlässigt das Detail. In der
Plastik ebenso wie in der Malerei. Furtwängler sagt über die
Gemmenplastik dieser Kultur: „Die mykenische Kultur kümmert sich
wenig um klare Technik und den Knochenbau des Körpers, darin die
ägyptische sich auszeichnet, aber die Wiedergabe des Fleisches,
der Muskeln ist ihr Element.“ Damit ist der impressionistische
Charakter dieser Plastik scharf gezeichnet, in der, ebenso wie in der
Malerei, die gesehene Form und nicht die Zeichnung Mittel der
künstlerischen Gestaltung ist. Das unmittelbare Festhalten des
Geschauten bewirkt, daß eine Figur wie dieser kretische
Gefäßträger so persönlich uns gegenübersteht, daß in den Gliedern
der Stiere und ihrer Bändiger auf Goldreliefs das Leben zu pulsieren
scheint, wie nie in einer ägyptischen Skulptur. Impressionismus und
Realismus haben sich zu allen Zeiten gegenseitig bedingt.
Abb. 9. Tongefäß aus Kreta.
Diese Stärke des Bildausdruckes macht die Figur körperhaft und
unabhängig von der Wand, an die sie gemalt ist. Die zeichnende
ägyptische Kunst geht, wie wir sahen, von der Wandfläche aus und
bildet ihre Gestalten flächenhaft, der kretische Impressionismus geht
von der Erscheinung der Gestalt aus und sucht die Form dann der
ruhenden Fläche einzugliedern, ohne daß das vollkommen gelingen
kann. Der Impressionismus hat zu jeder Zeit, in der er herrschte, den
Grundsatz l’art pour l’art — jede Kunst ist Selbstzweck —
proklamiert.
Am klarsten sprechen hier die Gebilde des Kunstgewerbes. Es
war schon in der Einleitung die Rede davon, daß das Kunstgewerbe
als Zweckkunst jedesmal ähnliche Stiltendenzen verfolgen müßte
wie die Architektur. In der Tat legt ein kretisches Gefäß (Abb. 9) den
Wert nicht auf die funktionelle Gliederung, sondern allein auf die
schöne Form. In wundervoller Ruhe gleitet hier die begrenzende
Linie vom Mund bis zur spitzen Endigung hinab. Gerade dieser
Abschluß beweist das Untektonische des Stiles. Denn ein struktiv
empfindender Stil bildet den Fuß des Gefäßes zum Feststehen aus,
während diese Spitze zwar einen wundervollen Zusammenfluß der
Linien gibt, aber stets einer besonderen Stütze bedarf. Wir werden
diese spitze Endigung bei fast allen untektonischen Epochen, in der
deutschen Spätgotik, im Rokoko wiederfinden, ebenso den
schmalen Hals, der die Linie des Gefäßbauchs in so ruhigem Fluß
weiterführt. Nur der Wulst, der den Hals abgrenzt, aber nicht
energisch genug, um die ruhige Überführung der Linien zu hemmen
und Hals und Körper wirklich zu sondern, gibt Kunde davon, daß ein
Gefühl für die Funktionen der Gefäßteile überhaupt vorhanden ist.
Ebensowenig wirkt der Schmuck gliedernd. Er geht nicht von der
begrenzenden Fläche der Gefäßwand aus, sondern von der
körperlichen Erscheinung seiner Motive, gliedert nicht das Gefäß,
sondern steigt an ihm auf. Man hat den Eindruck, vor einem
Aquarium zu stehen. Auf dem Meeresboden wachsen Korallen und
Seetang empor, die in der Strömung zu fluten scheinen,
Seeschnecken und Seesterne schwimmen zwischen ihnen, all das,
wie etwa in den Strähnen der schattenhaft sich wiegenden
Seepflanzen, von einer vollkommenen Formschönheit.
Es ist dasselbe Verhältnis wie in der Wandmalerei. Eine
Impression sagt hier in Schwarz und Weiß dasselbe, wie dort in
Farben, und wird durch Zurückführung auf einfachste Licht- und
Schattenflächen, ähnlich wie in japanischen Holzschnitten, der
Wandung eingefügt.
Abb. 10. Das Löwentor zu Mykenai.
Es ist einmal gesagt worden, daß der Unterschied zwischen den
Fürstensitzen auf Kreta und denen im gleichzeitigen Griechenland
derselbe gewesen sei wie zwischen einem Prunkschloß, wie
Versailles, und einer Burg, wie der Wartburg. Der Gegensatz ist
vollkommen. Während in Kreta die Gemächer und Vorratsräume sich
um offene Höfe gruppieren und miteinander durch eine Unzahl
Türen verbunden sind, geht die Tendenz auf dem Festland auf die
Klarstellung des einzelnen Gemaches. Das μέγαρον, der
Männersaal, ist der Zentralraum der Anlage. Er liegt vollkommen
isoliert, und auch die Nebengemächer liegen nicht Wand an Wand,
sondern an Korridore gereiht. Das einzelne Gemach hatte seinen
Eingang an der Schmalseite, so daß man es beim Eintritt
vollkommen überblickte, während in Kreta die Haupttür an der
Langseite lag. Und ebenso war die Gesamtheit der Anlage auf dem
Festlande energischer, wo der Mauerring den ganzen Komplex
zusammenhielt, als in Kreta, wo die Burgen Prunkpaläste waren, die
offen im Land lagen. Es ist in jeder Hinsicht der Kontrast zwischen
Zweckbau und Prunkbau, zwischen klarem Nebeneinander und
malerischem Reichtum. Naturgemäß sprechen die eigentlichen
Zweckbauten des Festlandes, die Befestigungsanlagen, diese Logik
am strengsten aus. Sie sind es, die als pelasgische oder kyklopische
Mauern so seltsame Legenden von Urweltriesen gezeugt haben.
Aber sie sind, wenn auch roh gefügt, in Wahrheit von großer
architektonischer Gewissenhaftigkeit. Das Löwentor von Mykenai
(Abb. 10), die wichtigste Architektur des Festlandes, ist ein
wohldurchdachtes Werk von gutem Ausdruck. Zwei Blöcke tragen
einen steinernen Balken, der aber selbst nur Last ist, von dem die
ganze Schwere der darüberliegenden Mauerschichten abgeleitet
und durch abgeschrägte Steinblöcke auf die seitlichen Mauern
übertragen ist. So bleibt über dem Tor ein dreieckiger unbelasteter
Raum, der mit einer leichten skulpierten Steinplatte gefüllt wird. Man
versteht also noch nicht, in Bogen zu wölben, aber es zeugt von
hoher architektonischer Geschmackskultur, daß gerade nur diese
Platte ausgeschmückt ist, die im Aufbau keine Funktion hat. Nun
weiß man sofort, welches Bauglied trägt, und welches getragen wird;
nur dem unbelasteten, füllenden Glied hat das tektonische Gefühl
hier dekorative Gestalt zu geben gewagt. So sind auch die
festländischen Geräte, deren Goldprunk die Burggräber von Mykenai
hergegeben haben, in ihrer Form mehr durch die Aufgabe bedingt,
als die kretischen. Sie sind geteilt nach der Funktion des Gerätes,
stehen mit breitem Fuß, nehmen mit geöffnetem Kelch den Wein auf.
Ihre Dekoration ist nicht einfach aus der Natur geholt, sondern zum
Gefäßornament geworden, das etwa als Saum wirkt. Zwar ist
Griechenland in der Epoche, in die die Burgen von Mykenai und
Tiryns weisen, d. h. etwa in der Mitte des 2. Jahrtausends v. Chr.,
stark von Kreta beeinflußt. Es gab damals dort Geräte mit lebhaften
Jagdszenen in landschaftlich tiefem Raum, die fraglos von Kreta
importiert sind. Auch die eigene Plastik, wie die Skulptur im
Entlastungsdreieck des Löwentors, spricht ihren Einfluß aus. Nicht
nur im Stil, der dem kretischen wesensverwandt ist und den Körper
in wundervoll fleischiger Rundung gibt; auch die Darstellung findet
sich im ganzen Kulturkreis, zwei Löwen, mit ihren Vorderpranken auf
eine Art Altar gestützt, zu Seiten der charakteristischen Säulenform
des Stils. Wahrscheinlich ist die Säule hier Symbol einer Gottheit, die
die Löwen bezwungen hat, zumal die Tiere gelegentlich wirklich an
die Säule gefesselt erscheinen. Dadurch ist eine zweite Deutung
weniger wahrscheinlich geworden, die wohl, als man die
Paralleldarstellungen noch nicht kannte, gerade durch das Löwentor
entstanden ist, daß nämlich die Säule das Haus symbolisierte, vor
dem die Löwen Wache hielten. Der Darstellung, die auch sonst oft,
vor allem auf geschnittenen Steinen, vorkommt, scheinen
orientalische Mythologieen zugrunde zu liegen. Jedenfalls ist die
kretisch-mykenische Kunst, die in dieser Vorzeit sich plötzlich mit
solcher Lebenskraft zutage drängt, nicht wurzellos gewesen.
Es muß freilich überraschen, hier, wo man in der Bronzezeit fast
an den Anfängen europäischer Kultur steht, eine ausgesprochene
Luxuskunst zu finden, deren Formen nicht aus dem Zweck geboren
sind, sondern, besonders in Kreta selbst, den klaren Zweckausdruck
geradezu negieren. Der Impressionismus im Fresko und in der
Dekoration der Vasen hat rein auflösende Funktion. Und wenn man
sieht, wie die Frauentracht damals den Körper mit einem glitzernden
Meer von ornamentierten Goldflittern umschimmerte, wider seine
Struktur in der Hüfte zusammenschnürte und die Brust hervorpreßte,
ja, wie die schlanke Taille auch beim Mann ein Schönheitsideal ist
(Abb. 8), so fühlt man unmittelbar den Reichtum des Rokoko als den
nächst verwandten Stil. Die hellenische Kunst, die auf demselben
Boden erwächst, ist in ihrem tektonischen Bewußtsein, das jede
Form ihrer Aufgabe gemäß ausgestaltet, der denkbar stärkste
Gegensatz.
Die Beziehungen zu Ägypten ermöglichen es, die Epoche
ziemlich sicher zu datieren. Schon ums Jahr 2000 v. Chr. muß der
Verkehr zwischen diesen beiden mächtigsten Anwohnern des
Mittelländischen Meeres begonnen haben. Unter dem Namen
Kaphtor kennt die Bibel ein Volk, das als Kephtiu huldigend auf
ägyptischen Reliefs erscheint, dieselben trichterförmigen Gefäße
tragend wie auf dem Fresko von Knossos. Man hat, wie wir sahen, in
Ägypten mykenische Stücke technisch und künstlerisch kopiert, und
auch die ägäische Kultur entnahm Ägypten manches Vorbild zu
selbständiger Verarbeitung. Aber um 1200 holen die Anwohner des
Ägäischen Meeres zu einem gewaltigen Schlage aus, unternehmen,
alle Kraft zusammenraffend, einen kriegerischen Vorstoß gegen das
Ägypterreich und werden zerschellt. Von da ab hören wir nichts mehr
von der ägäischen Kultur, und wenn wir dann sehen, daß die
Gefäße, die Hellas kurz darauf hervorbringt, ägäische
Überlieferungen in ärmlicher Entartung fortsetzen, dann haben wir
das Gefühl, einer der großen Tragödien in der
Menschheitsgeschichte gegenüberzustehen.
Welcome to our website – the ideal destination for book lovers and
knowledge seekers. With a mission to inspire endlessly, we offer a
vast collection of books, ranging from classic literary works to
specialized publications, self-development books, and children's
literature. Each book is a new journey of discovery, expanding
knowledge and enriching the soul of the reade

Our website is not just a platform for buying books, but a bridge
connecting readers to the timeless values of culture and wisdom. With
an elegant, user-friendly interface and an intelligent search system,
we are committed to providing a quick and convenient shopping
experience. Additionally, our special promotions and home delivery
services ensure that you save time and fully enjoy the joy of reading.

Let us accompany you on the journey of exploring knowledge and


personal growth!

ebookmass.com

You might also like