Machine Learning with Python: The Complete Guide

Learn how to apply linear regression within a machine learning workflow, using a mean squared error cost function to fit an optimal regression line with multiple features.

Understand how multi-feature linear regression uses learning rate and gradient descent to minimize mean squared error and optimize theta weights and intercept.

Learn to build a multiple linear regression model in Python using pandas and scikit-learn, with train_test_split for BMI-based features, and evaluate with predictions and mean squared error.

Learn to build a multiple linear regression model with machine learning, train-test split, and visualize predictions, using BMI and height, while exploring normalization and over fitting versus under fitting.

Apply linear regression with gradient descent to a new dataset, then visualize actual versus predicted values with a scatter plot, starting from easier problems and progressing to more challenging tasks.

Learn how to build a gradient descent based linear regression model, train and test with split data, and visualize actual versus predicted values across datasets.

Explore random forest basics and decision trees, including splitting, nodes, and pruning; learn the Gini index as a cost measure and interpret a simple iris dataset tree.

Build a simple decision tree and extend it to a random forest, using the Gini index to evaluate splits from weather, humidity, and wind in a golf example.

Build a random forest classifier with the iris dataset. Visualize a single decision tree using export graphic visualization, recording features and class names.

Visualize feature importance in a random forest with a bar graph, showing which features drive splits and how standard deviation reveals variance across trees, using the Iris dataset.

Build random forest models on a chosen dataset, compare forests with a scoring method to identify the best predictor, then create a bar plot of feature importances for that forest.

Explore how to train a random forest on a cancer dataset, compare trees to identify the best forest, and visualize feature importance with standard deviation bars.

Learn how support vector machines, a supervised algorithm, use kernel tricks to raise dimensionality and find optimal hyperplanes with a soft margin for clear class separation.

Explore how support vector machines form boundaries using a soft margin, focusing on the C and epsilon parameters, and learn how parameter tuning shapes the decision boundary in Python.

Train a support vector machine on the iris dataset, evaluate accuracy with train/test splits, and tune kernel types and hyperparameters such as gamma and C.

Visualize and compare support vector machine boundaries on the iris dataset using linear, poly, rbf, and sigmoid kernels, and explore how c and gamma influence boundary shapes and model accuracy.

Train a support vector machine on the wine dataset, score it across multiple trials, and visualize results to explore how random state and shrinking affect performance.

Explore naive bayes classifiers in text classification and see how Bayesian statistics use prior knowledge to inform likelihoods. Learn the Bayes formula and the role of prior, likelihood, and spam.

Explore Bayesian probability with practical spam and smoke-fire examples; learn priors, likelihoods, and applying naive Bayes to text processing and spam filtering.

Load and preprocess text data, perform data wrangling to convert to counts and tf-idf features, and build a Naive Bayes classifier to categorize science articles by topic.

Learn to build a naive bayes classifier for natural language processing tasks using tf-idf and count vectorizer, train on the 20 newsgroups data, and predict topics in new text.

Apply multi nominal naive bayes to classify news documents by superclass, massaging the dataset and labeling for science, politics, and more, and assess model accuracy.

Build a naive bayes classifier for a text dataset by massaging data, using count vectorizer and tf-idf, defining seven super categories like science and computers to achieve 68 percent accuracy.

Explore how to split data for machine learning training and testing, diagnose overfitting and underfitting, and apply k-fold and leave-one-out validation to improve model generalization.

Explore cross-validation techniques, including leave-one-out and k-fold, to split data for training and testing. Learn how these methods reduce overfitting and underfitting while improving model generalization.

Learn to implement k-fold and leave-one-out cross validation using a support vector machine on the iris data, with train-test splits and accuracy assessment.

Explore how to deliberately induce overfitting and underfitting in machine learning models using the iris dataset, then analyze causes and remedies to improve future performance.

Explore how overfitting and underfitting arise in an iris dataset using a support vector machine with a radial basis function kernel, visualize boundaries, and adjust train-test splits to improve generalization.

Explore the k nearest neighbors algorithm as the final supervised method, contrast it with k means clustering, and examine distance metrics such as Manhattan, Euclidean, and Koski distances.

Explore distance metrics in machine learning, including the Murkowski distance, and compare Euclidean and Manhattan distances, applying them to the k-nearest neighbors algorithm.

Explore k-nearest neighbors classification with Euclidean and Manhattan distances using the iris dataset, visualize decision boundaries, and compare cross-validated accuracy and the p parameter effects.

Explore k-nearest neighbors regression on the iris data to predict petal width from length and width, comparing scaled versus unscaled inputs and evaluating mean squared error with cross-validation.

Explore how k nearest neighbors can serve as regression, compare regression and classification targets, apply a uniform weight distribution with Manhattan distance; note iris is unsuitable for regression.

Learn the difference between regression and classification by building a KNN regressor on diabetes dataset with three neighbors and cross-validation, noting the Iris dataset cannot be used for this regression.

Explore unsupervised learning with k-means clustering, which groups data into clusters without labels using distance metrics like Euclidean or Manhattan.

Explore hierarchical k-means clustering, starting from single points and merging them into two or more clusters using distance metrics, dendrogram visuals, and practical iris dataset examples.

Build a k-means clustering model using iris data to compare actual labels with predicted clusters. Explain unsupervised learning differences from supervised learning, and discuss parameters like initialization and iterations.

Explore hierarchical k-means clustering in Python by building a dendrogram and visualizing how agglomerative linkage merges create three clusters, using nearest neighbors connectivity and ward linkage.

Explore Markov processes and hidden Markov models, learn how transition probabilities connect states, and see how discrete states and latent variables drive unsupervised learning.

Explore hidden Markov models and latent features that influence state transitions and emissions, using weather and urn examples to show how emission and transition probabilities shape observable sequences.

Learn to code a simple two-state hidden Markov model using the hmlearn package, install via conda, initialize start probabilities and transition matrices, fit data, and generate samples.

Explore Gaussian mixture models and normal distributions, learn how expectation maximization selects means and variances, compare GMMs to k-means, and extend from single to multi-feature data.

Explore multivariate gaussian mixture models with two features, learning to merge normal distributions, estimate means and variances, and use expectation maximization to maximize joint probability.

Build a multivariate gaussian mixture model and review its parameters, while exploring cold start and warm start issues, k-means initialization, and expectation maximization for effective learning.

Visualize gaussian mixture model boundaries with a plotting library, rendering ellipses for each component by means, variances, and weights. Explore how training parameters shape the boundaries.

Explore gaussian mixture models by building a base gmm and compare it to bayesian gaussian models, reading about bayesian approaches and highlighting how bayes gmm differs from standard gm.

Explore Bayesian Gaussian mix t ure models and how prior knowledge influences inference, training, and visualization compared with traditional Gaussian mixtures; build, fit, and evaluate with data blobs.