comparing-classifiers

This project compares various classifiers:

  1. K-nearest neighbour
  2. Bayes classifier
  3. Naive Bayes classifier

for the synthetic data belonging to one of the 3 types: (i) Linearly seperable (ii) Non-linearly seperable (iii) Overlapping alt text

Few experiments are also done on a real-world dataset, which can be a mixture of all the categories considered above. Further, prior distributions considered with the family of Bayes classifiers are gaussian and gaussian mixture model (GMM).

Setup info

  • For running the KNN classifier, simply run KNN.m
  • For the GMM task on a given dataset, run gmm.m
  • For the GMM task on real-world datasets, run gmm_real_world_static_data.m
  • For Naive Bayes task, run Naive_Bayes.m
  • RunThisFile.m - for Bayes and Naive Bayes with Gaussian distribution. Run this in the end.

Conclusions

Project_Report.pdf contains a detailed report about the performance analysis of the above classifiers on different datasets.

Authors

Madhura Pande

Aakriti Budhraja

Sadbhavana Babar