Cs434 Assignment 1
Due: Monday Oct 13th in class
Part I. (14pts)
Through out the course, Weka will be a very useful tool for you to explore different
machine learning and data mining algorithms. The purpose of this part of the assignment
is to familiarize you with this software package. For this assignment, please download
and install the Weka software package (version 3.4) from
http://www.cs.waikato.ac.nz/ml/weka/
If you don’t have access to a computer for this purpose, please inform the instructor and
special arrangement can be made to accommodate your need.
Basic information on this software package can be found in the following tutorial:
http://easynews.dl.sourceforge.net/sourceforge/weka/ExplorerGuide-3.4.pdf
Note that there are many documents available on the Weka webpage introducing different
aspects of Weka. For example, another useful thing to read is the following document,
which describes the input format for Weka, i.e., the “arff” (attribute relation file format)
format.
http://www.cs.waikato.ac.nz/~ml/weka/arff.html
Please use Weka to explore the “iris” data set that comes with the software. To open this
data set, choose “explorer” from the Weka GUI chooser, which opens a panel with
several tabs. Select the “preprocess” tab and click “Open file”, then click on the “data”
folder, choose the “iris.arff” file.
With the help of the Weka software, answer the following questions:
1. How many classes there are in this data set? (2pts)
2. How many (non-class) attributes there are? (2pts)
3. What are the mean and standard deviation for each attribute? (2pts)
4. If you were to choose only one attribute to build your classifier, which attribute
should you choose? (4pts)
5. Which pair of attributes provides the best discrimination among classes? (4pts)
(Suggestion: use the visualization tool to look for good separations among classes.)
Note that questions 4 and 5 are subjective; please provide your reasons for the answers.
Reasons could be describe in words and/or shown through figures.
