Machine Learning In-Depth (With Python)

Explore the foundations of AI, machine learning, and deep learning; compare supervised and unsupervised methods with iris, spam filter, and house price examples, and cover training, testing, and deployment.

Explore end-to-end machine learning, from data exploration to modeling and monitoring, covering supervised and unsupervised tasks like regression, classification, clustering, and dimensionality reduction.

Explore end-to-end ml with regression and classification, key metrics (rmse, mae, r-squared, adjusted r-squared), confusion matrix, accuracy, precision, recall, f1, and Titanic examples.

Get data from sources, perform exploratory data analysis, clean and wrangle, prepare features with one-hot encoding and scaling, and evaluate models with train-test splits, cross-validation, and hyperparameter tuning.

Analyze overfitting and underfitting in linear and polynomial regression, using train-test splits, RMSE, and regularization (ridge and lasso), while addressing multicollinearity, PCA, and gradient descent.

Analyze feature selection for linear regression using forward and backward stepwise methods, and apply recursive feature elimination to choose the best subset; cover ridge, lasso, elastic net, and scaling.

Explore a Kaggle house price prediction use case using linear regression, applying lasso, ridge, and elastic net within a crisp-dm workflow, including data preparation, modeling, evaluation, and Mlflow deployment.

Explore the shift from regression to classification, with real-world examples like spam filtering, iris, Titanic, telecom churn, and Mnist; learn confusion matrices, accuracy, precision, recall, and F1 trade-offs.

Explore how a decision tree handles classification and regression on nonlinear data, building interpretable rules with cart algorithm, Gini impurity, and information gain, illustrated by iris data.

Explore how a decision tree, a non-parametric model, builds leaf nodes and splits data with conditions. Tune depth and sample-related hyperparameters with grid search CV to control overfitting.

Explore ensemble learning with random forest, building multiple decision trees via bagging and feature sampling, understand out-of-bag evaluation, and balance bias and variance for robust predictions.

Explore support vector machines (SVM) and their use for classification and regression, including high dimensional spaces, max margin and soft margin, with kernel tricks for linear and non-linear data.

Learn dimensionality reduction through principal component analysis, including why it preserves variance, how to choose components, and practical demos with MNIST, plus incremental and kernel PCA and hyperparameter tuning.

Apply principal component analysis for dimensionality reduction before a support vector classifier, tune the number of components with grid search and five-fold cross-validation, and evaluate performance on Titanic data.

Explore unsupervised learning with clustering, including K-means, hierarchical, and DBSCAN, and learn how to form meaningful customer and image segmentation using centroids and evaluation methods.

Explore a customer dataset through data exploration, cleansing, feature engineering, and scaling, then apply k-means clustering to reveal two primary customer segments and insights for targeted marketing.