This is a simple implementation of logistic regression for classifying 2D points. ----------------------GETTING STARTED---------------------- 1. Press `g` to generate some points 2. Press `t` to train the model 3. Press `l` to see the loss plot and then `d` to revert back to see points 4. Press `n` and then `e` to evaluate the model on noisy points ------------------------------------------------------------ ------------------------CONTROLS------------------------ - g: generate points - t: train the model - n: generate noisy linear points - e: evaluate current model on current existing points - p: preprocess data - l: view loss graph - d: draw points and plots (revert from loss view) - r: reset all - u: reset weights only - 1: set to generate linearly separated points (press `g` after) - 2: set to generate quadratic separated points (press `g` after) - 3: set to generate circle separated points (press `g` after) ------------------------------------------------------------ ------------------------NOTES------------------------ - The model only learns two single value weights and one bias term - Since the model only uses two weights, it does not show an estimate plot for decision boundaries other than linear ------------------------------------------------------------
Logistic Regression explained here https://www.youtube.com/watch?v=hjrYrynGWGA&feature=youtu.be
