Linear, Logistic and Ridge Regression

Linear Regression

link to repository

Dataset from linRegData.npy. The data is a matrix (100, 2). Column 1 is $x$ and Column 2 is $y$. Data is be read using $numpy.load(’linRegData.npy’)$. Given a function $y = f(x)$, using linear regression a program to find the line which best fits the data.

Ridge Regression

link to repository

Dataset from linRegData.npy. The data is a matrix (100, 2). Column 1 is $x$ and Column 2 is $y$. Data can be read using $numpy.load(’linRegData.npy’)$. For this exercise, fit a polynomial of degree 15 to the data using ridge regression. I.e. $x$ is converted to $[1, x, x^2, x^3, . . . , x^{15}]^T$. Using 5-fold cross validation, estimating the best $λ$ from the set, $λ = [0.01, 0.05, 0.1, 0.5, 1.0, 5, 10]$

Logistic Regression

link to repository

training a basic logistic regression classifier to classify two set of digits from the MNIST dataset. After preprocessing, the dataset loaded contains only digits 0 and 1. logReg.py implements logistic regression. The below plot shows training error vs. iterations.