Introduction to ML Classification Models using scikit-learn

Clarify what machine learning is and when to use it, contrasting it with rule-based methods through a spam classification example, and outline training on labeled data.

Explore supervised and unsupervised learning, distinguishing classification and regression, including linear regression, feature vectors and x and y variables, labels, and the least squares method.

Explore how the support vector machine, a binary classifier, uses a linear hyperplane to separate spam from ham and how transforming data to a higher dimensional space enables nonlinear separation.

Learn to build a support vector machine classifier using the adult dataset to predict whether annual income exceeds $50,000, with data loading, preprocessing, and visualization in Python.

build and tune a support vector classifier in scikit-learn using train test split, with features like education, occupation, age, and gender, plus correlation heatmaps, achieving about 80 percent accuracy.

Learn how decision trees classify outcomes by splitting inputs from root to leaves, using feature vectors and labels learned through recursive partitioning in supervised learning with scikit-learn.

Learn how decision tree learning uses recursive partitioning to split housing data by type, rent per bedroom, and yard, producing transparent, binary city predictions for Mumbai or Bangalore.

Explore information gain and entropy in decision trees for classification, compare information content across outcomes, and note Gini impurity as used by scikit-learn.

Build a decision tree classifier to predict wine quality using a white wine dataset; load csv data, explore correlations, split data, train a decision tree, and evaluate accuracy.

Explore how to view and tweak a scikit-learn decision tree, inspect feature importances and alcohol content splits, and experiment with max depth, max features, and information gain versus Gini impurity.

Explore overfitting in machine learning, contrasting training and unseen data, and learn mitigation techniques such as cross-validation, regularization, and ensemble learning.

Explore how cross-validation and regularization mitigate overfitting by balancing bias and variance, with data splits, multiple models, and feature combinations.

Explore ensemble learning as a strategy to mitigate overfitting by combining results from multiple models. Discover how random forests leverage diverse decision trees and majority voting to improve predictions.

Build a random forest classifier by ensembling decision trees with the Portuguese wine quality dataset; reuse decision-tree code and tune n_estimators, max_depth, and max_features to compare with a single tree.