AI Bootcamp: Beginner to Expert in Machine Learning 2024

Learn how machine learning lets computers learn from data to predict benign or malignant cells, using cleaned data and predictive models, and explore key techniques from regression to recommendations.

Learn to use Python for machine learning with key libraries NumPy, SciPy, Matplotlib, Pandas, and scikit-learn, building, evaluating, and exporting models through a hands-on, lab-focused workflow.

Compare supervised and unsupervised learning, understanding labeled data for classification and regression, and discover how clustering, dimensionality reduction, and feature selection reveal attributes and structure.

Explore how regression uses historical car data to predict continuous outcomes like CO2 emissions, distinguishing dependent and independent variables and simple versus multiple regression.

Explore how simple linear regression uses a fit line to link engine size and CO2 emissions, then predict emissions for new cars.

Learn regression model evaluation using training on the same data, train-test splits, and k-fold cross-validation. Understand training versus out-of-sample accuracy and how ground truth guides model validity.

Assess regression models by comparing real values to predictions with average absolute and squared error. Explain root average squared error, relative absolute error, relative squared error, and r².

Explore multiple linear regression, using several predictors to predict a target like CO2 emissions, and learn how to minimize mean squared error with ordinary least squares or gradient descent.

Explore classification, a supervised learning method that uses labeled training data to assign test cases to categorical outcomes, enabling binary and multi-class predictions across finance and recognition tasks.

Explore the k-nearest neighbors (KNN) classification method, using age and income to assign customers to groups via distance-based voting, majority vote, normalization, and careful K selection to avoid overfitting.

Learn how to evaluate classification models using metrics like Jacquard index, confusion matrix, precision, recall, f1 score, and log loss, illustrated with a churn dataset.

Learn how decision trees classify data and predict drug A or B using age, gender, pressure, blood pressure, and cholesterol by splitting data with attribute tests.

Build decision trees through recursive splitting to reduce node impurity and maximize information gain. Explore how entropy and disorder guide feature selection, comparing sex and cholesterol to achieve purer leaves.

Explore logistic regression as a classification method, predicting churn and other binary outcomes using features like tenure, age, and income. Learn about probability estimates and when to use logistic regression.

Compare linear and logistic regression, showing linear regression's limitations for binary tasks, and introduce the sigmoid function to produce class probabilities for logistic regression.

Train a logistic regression model by adjusting parameters with the logistic cost function and gradient descent. Minimize the cost to accurately predict churn.

Explore how support vector machines classify data by mapping to higher dimensional spaces with kernel functions, select the maximum-margin hyperplane using support vectors, and apply to image and text analysis.