Machine Learning Experiments with WEKA or R Studio

These Machine Learning Experiments are according to Pune University BE IT Syllabus (SL3).

Guidelines for Students and Teachers: Experiments should be performed with WEKA or R. Students are also encouraged to implement the experiments with Java 1.6 and higher version (RJava Package). Standard Data Sets available on line may be used. A few popular data sets are : 1) Olive Oil Data Set 2) Iris Data Set 3) UC Irvine ML Laboratory
REFERENCE : 1) Open source software-WEKA or R 2) JAVA 6.1 or more ( for RJava Package) Subject teachers are advised to frame proper assignment statements from the following list.


  1. Study of platform for Implementation of Assignments Download the open source software of your interest. Document the distinct features and functionality of the software platform. You may choose WEKA or R or Rjava.
  2. Supervised Learning - Regression Generate a proper 2-D data set of N points. Split the data set into Training Data set and Test Data set. i) Perform linear regression analysis with Least Squares Method. ii) Plot the graphs for Training MSE and Test MSE and comment on Curve Fitting and Generalization Error. iii) Verify the Effect of Data Set Size and Bias-Variance Tradeoff. iv) Apply Cross Validation and plot the graphs for errors. v) Apply Subset Selection Method and plot the graphs for errors. vi) Describe your findings in each case.
  3. Supervised Learning - Classification Implement Naïve Bayes Classifier and K-Nearest Neighbor Classifier on Data set of your choice. Test and Compare for Accuracy and Precision.
  4. Unsupervised Learning Implement K-Means Clustering and Hierarchical clustering on proper data set of your choice. Compare their Convergence.
  5. Dimensionality Reduction Principal Component Analysis-Finding Principal Components, Variance and Standard Deviation calculations of principal components.
  6. Supervised Learning and Kernel Methods Design, Implement SVM for classification with proper data set of your choice. Comment on Design and Implementation for Linearly non separable Dataset.


Post a Comment