The Weka Explorer
Due Wednesday 17 April 2013
The goal of this week's lab is to learn how to use the Weka Explorer interface to the Weka machine learning tools.
- Go through the tutorial in the Data Mining text, chapter 17 sections 1-3.
Download the Bird Arrivals
Summary data file. Use the Weka Explorer to see if you can
predict whether a bird is a warbler based on its arrival pattern using
a decision tree, and at least three other classifier methods.
For each case, analyze the resulting classifier and discuss what it is doing in your writeup. Diagrams help here.
Note that the types row has been deleted from this version of the CSV file because it confuses Weka.
Alternatively, you may use another data set of your choice where you are trying to predict one variable based on multiple other independent variables. Please talk with the prof before you take this option.
- Using your program from last week, try clustering the Bird Arrivals Summary data with between 5-10 clusters and summarize your results (this version has the types row). Do the groupings make sense to you?
- Try running PCA on the bird arrivals summary data. What do you find?
- After running PCA, try clustering again and show the results plotted on the first two eigenvectors.
- Try out some other data sets with Explorer.
- Try out other capabilities of Explorer, such as clustering.
For this week's writeup, create a brief wiki page that shows screenshots of your results from Explorer. For each screen shot briefly explain what they mean, and what they show.
Once you have written up your assignment, give the page the label:
There is no code to handin this week unless you wrote code specifically to do an extension.