navreeetkaur / machine-learning-algorithms

Implementation of Machine Learning algorithms from scratch

Machine Learning Algorithms

These implementations were done as part of assignments of the Machine Learning course ELL409(2018-19) @ IIT Delhi

This project consists of the tasks of building up different kinds of classifiers on various example datasets. PFA : Detailed report for data analysis on FMNIST, Railway Data, Health Data and River Data.

We have implemented following types of Classification schemes for each problem

1. Bayes Classifier (with different class conditional densities and estimation techniques)
2. Naive Bayes Classifier
3. K-means Clustering
4. K-Nearest Neighbor Classifier
5. Principal component analysis (where ever applicable)
6. Linear Models for Binary Classification, Regression and Multi-Class Classification
7. Logistic Models for Binary and Multi-Class Classification
8. Binary and Multi-Class FLDA
9. Perceptron ALgorithm
10. Support Vector Machine

machine-learning-algorithms's Issues

Move all code to a directory named src

PCA does not perform as good as scikit implementation

Perceptron not as good as scikit implementation

Integrate LDA and SVM to main Classifier file

Add a fixed seed to uniform sampling of test train split for railway & river

Change to inputting file paths rather than names

Gradient Descent Approach to Multi Class Linear Classification

Not working as of yet. Does not converge