Machine Learning Mastery: From Basics to Advanced Techniques

Explore the supervised learning workflow in Python, covering data collection and preparation, feature engineering, model selection, training, and evaluation using the iris dataset.

Learn to evaluate model performance with train-test splits, measure accuracy, and explore overfitting and underfitting while visualizing how model complexity affects results using k-nearest neighbors approach and the iris dataset.

Learn to create features, fit a linear regression model in Python, and visualize predictions to gain insights.

Explore the fundamentals of linear regression, including fitting and predicting with models. Evaluate performance with mean squared error and R squared, visualize results, and implement with numpy, matplotlib, and scikit-learn.

Explore cross-validation for regression, implement five-fold r-squared scoring with scikit-learn in python, and analyze cross-validation metrics to assess model generalization.

Explore regularized regression to improve linear models by adding penalty terms, compare ridge and lasso for feature importance, and evaluate with mean squared error on the Boston housing data.

Explore the k nearest neighbors classification method, training on the iris dataset and predicting with k=3, then evaluate accuracy and discuss choosing the right k.

Decide a primary evaluation metric and assess a diabetes prediction classifier using metrics like accuracy, precision, recall, F1 score, and AUC ROC with a random forest and train-test split.

Learn to build a logistic regression model in Python for binary classification, visualize the ROC curve, and use ROC AUC to assess model discrimination and performance.

Explore hyperparameters and master grid search with GridSearchCV and RandomizedSearchCV to optimize a random forest on the breast cancer dataset.

Explore data preprocessing essentials by creating dummy variables for categorical features and applying them to linear and logistic regression models, evaluating with mean squared error.

Learn techniques for handling missing data with practical strategies and build a robust machine learning pipeline for income prediction. Explore imputation, scaling, and a random forest classifier.

Learn how centering and scaling in data preprocessing improve regression and classification models, demonstrated with scikit-learn linear regression and random forest on breast cancer and Boston datasets.

Evaluate multiple models to compare regression and classification performance using visualizations and test set predictions, and build a pipeline for predicting song popularity.

Refresh your scikit-learn knowledge with a practical KNN classification on iris and synthetic data, and focus on model comparison and overfitting mitigation.

Explore logistic regression and support vector machines for text classification and sentiment analysis of movie reviews using scikit-learn, tf-idf features, and the imdb dataset.

Explore linear classifiers and linear decision boundaries, visualize decision boundaries with logistic regression on synthetic data, and understand how these models make decisions across simple to complex scenarios.

Explore how linear classifiers use coefficients and intercept to make predictions, visualize the decision boundary with logistic regression, and observe how altering coefficients reshapes model behavior.

Explore loss functions as metrics guiding model evaluation and optimization, focusing on zero-minus-one loss and how minimizing the loss improves accuracy, illustrated with linear regression.

Explore loss function diagrams for classification, comparing logistic and hinge losses, and implement logistic regression to visualize decision boundaries and assess model accuracy.

Explore logistic regression beyond classification by obtaining class probabilities, examine how regularization affects these probabilities, and visualize easy and difficult examples to reveal model confidence.

Explore multi-class logistic regression, count coefficients, fit the model on the iris dataset, visualize decision boundaries, and compare with a one-versus-rest SVM approach.

Explore unsupervised learning through clustering 2D data, using the elbow method to determine optimal clusters, and evaluate results with inertia and silhouette analysis.

Explore how to determine the optimal number of clusters for grain data in an unsupervised learning setting using k-means and the silhouette score, with data standardization.

Transform feature representations and scale data to improve clustering with k-means, demonstrated on fish data and stock movements.

Explore hierarchical clustering and the concept of how many merges, visualize hierarchies with dendrograms, and apply the technique to digits data and stock datasets.

Leverage t-SNE to visualize high-dimensional data in a two-dimensional map, uncovering patterns and relationships in datasets and in the dynamic stock market.

Explore dimension reduction with PCA, applying principal component analysis to fish measurements, tf-idf word frequencies, and clustering Wikipedia articles to streamline high dimensional data.

Explore non-negative matrix factorization to extract interpretable topics from documents and parts from images. Compare nmf with pca to see why nmf reveals meaningful parts in data.