Homeworks

HW1: Using the SMS Spam Collection dataset, write a small program (Python?) that implements a Naive Bayes spam classifier from scratch. Your program should calculate the prior probabilities of spam and non-spam messages, estimate the probability of selected words appearing in each class, and use Bayes’ rule to classify new messages as spam or non-spam. Assume that the words in a message are conditionally independent given the class. Test your classifier on a small set of messages and print the predicted class and probability for each message. Do not use a machine-learning library or an existing Naive Bayes implementation; implement the probability calculations yourself.

HW2: Using the Breast Cancer Wisconsin (Diagnostic) dataset, write a small program (Python?) that implements logistic regression from scratch to predict whether a tumor is benign or malignant. Your program should select two or three numerical features from the dataset (don't select all), initialize the model weights and bias randomly, calculate the predicted probability using the sigmoid function, and use gradient descent to update the weights and bias over multiple training iterations. After training, use the model to classify the test data and report the classification accuracy. Do not use a machine-learning library or an existing logistic regression implementation, etc.

© 2026, Particle