Weka Data Mining Software Installation and Processing

There are two steps in this assessment which prepares you for identification and analysis work you are expected to complete throughout this subject. The first task will build the practical and technical skills that will enable you to compare and evaluate output patterns for visualization.

For this first assignment, you are required to install the Weka software, that you will then use throughout the duration of this subject. You will also do some tasks with the Weka software for data preprocessing and visualization.

Task 1: Data Preprocessing

Load a dataset by clicking the Open file button in the top left corner of the panel. Inside the data folder, which is supplied when Weka is installed, you will find a file named cpu.with.vendor.arff.

As shown in the Weka interface, the weather data has 209 instances, and 8 attributes called vender, MYCT, MMIN, MMAX, CACH, CHMIN, CHMAX, class. Click on the name of an attribute in the left subpanel to see information about the selected attribute on the right, such as its values and other details. This information is also shown in the form of a histogram. Answer the following:

(a) What are the values that the vender attribute has? [2 marks]

(b) What is the class value of instance number 80 in the cpu.with.vendor data? [1 mark]

(c) Load the cpu.arff dataset and open it in the editor by clicking the Edit button from the row of buttons at the top of the Preprocess panel in Weka Interface and answer the following question. How many numeric and how many nominal attributes does this dataset have? [2 marks]

Task 2: Visualization and Analysis

Load the iris.2D.arff dataset in Weka and answer the following questions.

(a) How many instances and how many attributes does this dataset have? [1 mark]

(b) What is the range of possible values for each of the 2 attributes that can be observed in the dataset? [2 marks]

(c) Present a scatter plot visualization of this dataset and find which two classes have more overlapping tendency and which one is likely to be a separate class as observed in the attribute-pair based plotting. Alternatively, you may use the 3D visualization feature provided in Weka to find which two classes have more overlapping tendency and which one is likely to be a separate class using different combinations of any three featuring attributes out of four attributes in the dataset. [2 marks]