alexmainstone/masters-acceptance-ml

Implementation of 3 gradient descent algorithms. Using no libraries.

Linear descent

Simply run train-model.py to train & test-model.py to test

Gradient Descent

The code will train & test 3 models of gradient descent:

• Standard Gradient Descent
• Stochastic Gradient Descent
• Minibatch Gradient Descent

Code

• model.py - Contains the model code, and each gradient descent algorithm
• utils/csvreader.py - reads the csv data into python lists
• utils/standardscaler.py - applies data = (data - mean) / std for each feature
• train-model.py - trains the models
• test-model.py - tests the models

Includes

The pickle library is used to serialize the models into a models folder.