Implemented following models-
- Binary Bayes Classifiers from data with Max. Likelihood.
- Multiclass Bayes Classifiers from data with Max. Likelihood.
- Bias-Variance analysis in regression.
- Analyse overfitting and underfitting in Regression.
Implemented following models-
- Implemented Logistic Regression with RBF kernel and did the hyperparamter tunning(kernel parameter, regulariser, learning rate).
- Implemented SVM for different kernel type(RBF, linear, poly).
- Implemented Decision tree with purning(at what height stop spliting).
- Implemented Random Forest Classifier and tuned with(hyper parameter- Fraction of data to learn tree=0.5, Fraction of number of features taken per data=0.5).
Implemented following models-
- Run k-Nearest neighbours on the binary classification dataset for classifiying whether a given movie is a comedy or not.
- Implemented PCA and regression.
- Implemented Baseline methods for collaborative filtering.
- Implemented EM algorithm for Gaussian Mixture models.
Implemented following model-
- Implemented Guassian Mixture Model.
- Implemented PCA, KPCA.
Implemented different models for classification of ham-spam mails.
In this contest we had to predict the movie rating, the famous netflix challenge.
we tried different collaborative filtering method. Ex-Nearest neighbourhood models, Modified latent factor model.
Our best result came with Modified latent factor model.
Contest was hosted on kaggle https://www.kaggle.com/c/prml19/leaderboard .