Assignments of the course

Implemented Linear and Logistic regressions from scratch using only NumPy.

Logistic regression implemented using Sklearn.

SGD implementation using Sklearn

Conducted EDA on Bank Authentication Dataset (in datasets folder)

Implemented SkLearn's PCA, SVD, Tsne. Did a stratified split and trained a logistic regression using Sklearn.

Used Bootstrapping on the Linear regression model to measure bias and variance of the model.

• Implemented Gridsearch and K-cross validation from scratch and used it on Sklearn's Decision-Trees and GNB classifier.
• Calculated Confusion matrix for binary and multiclass data and plotted their ROCs from scratch

Implemented Gaussian Naive Bayes Classifier from scratch using only NumPy library

Implemented Multilayer Perceptron with parameters: -

• n layers: Number of Layers (int)
• layer sizes: an array of size n layers which contains the number of nodes in each layer. (array of int)
• activation: activation function to be used (string)
• learning rate: the learning rate to be used (float)
• weight init: initialization function to be used
  • zero: Zero initialization
  • random: Random initialization with a scaling factor of 0.01
  • normal: Normal(0,1) initialization with a scaling factor of 0.01
• batch size: batch size to be used (int)
• num epochs: number of epochs to be used (int)

Used the MLP from Q1 on the MNIST dataset for various hyper-parameter settings

Used PyTorch's MLP for a dataset.

Used pre-trained Alexnet model from PyTorch.