Machine Learning from Scratch

Build a simple linear regression model from scratch using manual calculations with y = mx + c, then predict taxi fares and assess fit with mae and rmse.

build a simple linear regression model in Python on Google Colab to predict outputs from inputs and fit the regression line, using mean squared error and r square for evaluation.

Learn to solve multiple linear regression manually with a three-equation model using x, y, z to predict force. Validate with testing data and assess accuracy with RMSE and R-squared.

Learn how logistic regression handles binary classification by using the sigmoid function to produce probabilities and a 0.5 threshold for decisions. Apply to spam detection, medical diagnosis, and pass/fail tasks.

Compare linear and polynomial models by fitting curves with polynomial regression, and form equations from data through curve or equation fitting to predict unknowns, using RMSE as a metric.

Learn to apply multiple linear regression in Python to predict cutting force from feed speed and depth of cut, using a CSV dataset with a train-test split and evaluation metrics.

Explore the k nearest neighbor algorithm, a lazy supervised learner for classification and regression, using distance metrics like Euclidean distance, Manhattan distance, and cosine similarity, with cross-validation to select k.

Explore the random forest algorithm, an ensemble of multiple decision trees that reduces overfitting and improves accuracy, with outputs combined through majority voting or averaging for classification and regression.

Explore how the random forest algorithm builds multiple decision trees, uses ensemble voting to produce final predictions, and applies bagging and boosting for classification and regression.

Learn how the random forest regression model uses multiple decision trees to predict vehicle mileage from rpm and load, by grouping data and averaging predictions.

Evaluate models with classification metrics such as accuracy, precision, recall, and F1. Also use regression metrics like MAE, MSE, RMSE, and R-squared to assess numerical predictions.