Hands-On Machine Learning: Python Project Showcase

Explore real-world machine learning case studies across regression, time series, clustering, and classification. Apply statistics, probability, and data visualization to build practical data science solutions.

Set up the deep learning environment by installing Python and the Anaconda framework, launching Jupyter notebooks in the browser, and validating versions across platforms.

Analyze the problem statement of linear regression using datasets such as R&D spend and salary data, and apply y = a x + b to model profit and salary.

Launch a jupyter notebook to run normal linear regression on salary data csv, using years of experience to predict salary, with train/test split and a visual regression line.

Explore polynomial regression using the position salaries.csv dataset, implementing polynomial features with degrees 2 and 3 alongside linear regression, and visualize the results.

Apply backward elimination on a 50 startups dataset, select profit and spending features (R&D, administration, marketing), encode state with get dummies, and visualize a linear regression.

Explore robust regression in a Jupyter notebook by generating x and y data with outliers, visualizing with matplotlib, and comparing linear regression, OLS, and Theil-Sen regressor using NumPy and scikit-learn.

Compare linear and logistic regression, highlighting that logistic regression targets discrete outcomes and outputs probabilities, then implement logistic regression on a Titanic-like dataset using pandas, numpy, and visualization tools.

Visualize logistic regression through count plots of survived by pclass and sex, analyze age with distance plot, explore fare with histograms, and display heat map before introducing k-means clustering.

Explore k-means clustering in unsupervised learning using two datasets, including salary and turnover data, to identify similar groups and address two major clustering problem statements.

Learn to implement a Euclidean distance calculator and apply it to k-means clustering by computing distances between data points and random centroids, updating centroids, and visualizing results with scatter plots.

Display k-means centroid values by setting the number of clusters, importing kmeans, and printing the resulting cluster centers and labels from a sample dataset.

Run the Python execution of the dataset to analyze face detection, observe the dataset folder with 100 images, and see the recognized faces as you advance to the next module.

Analyze the problem statement for building a time series from airline passenger data, explore 1949–1960 monthly trends, and prepare, train, and evaluate a keras-tensorflow model with plotting.

Develop a time series model using an LSTM in Python, importing NumPy, pandas, and Keras, applying min-max scaling with sklearn preprocessing, creating look-back datasets, and evaluating with mean squared error.

Learn to build a time series model by normalizing data with min-max scaling and splitting into train and test sets. Configure a stateful lstm with lookback for predictions.

Develop and evaluate invert predictions using inverse transforms, compute RMSE for train and test, and plot time series forecasts with lookback features from an lstm model on airline passenger data.

Analyze a Bitcoin time series for the first week of August 2017 by plotting the closing price in USD, using NumPy, pandas, matplotlib, and seaborn to visualize the date-indexed data.

Classify fruits data set from fruits with colors dot txt csv, using mass, width, height, and color score, then build train and test sets and apply logistic regression in Python.

Explore fruit type distribution through group-by counts and seaborn plots, including count plots, box plots, histograms, and scatter matrices to visualize mass, height, and color score.

Create training and testing data with train_test_split, split into x_train and x_test using random_state=0, scale with min max scaler via fit_transform on x_train and transform on x_test, evaluate logistic regression.

Explore logistic regression using scikit-learn, fit the model on training data, and evaluate accuracy on both training and testing sets, then preview a move to decision trees.

Train and compare decision tree classifiers on fruit subtype data, adjust max_depth, and report training and testing accuracy.

Demonstrates building a k nearest neighbors classifier, importing KNeighborsClassifier, fitting on training data, and evaluating accuracy on training and testing sets using score outputs.

Apply linear discriminant analysis in Python with sklearn to fit training data, evaluate accuracy on training and testing data, and compare with logistic regression in machine learning case studies.

Apply Gaussian Naive Bayes using sklearn to train on x_train and y_train, evaluate 0.86 training accuracy and 0.67 test accuracy, and prepare for plotting the decision boundary.

Plot the decision boundary of a k-NN classifier on a fruit dataset using mass, height, and color score, with train-test split, color maps, and a mesh grid.

Plot decision boundary by building a NumPy mesh grid for x and y values, predicting with a decision tree, and rendering a mesh that distinguishes apple, mandarin, orange, and lemon.

Analyze credit card defaulters from an xlsx dataset renamed default_credit_card, examining limit_balance, sex, education, payout, and bill amounts.

Prepare a Python dataset from the default credit card Excel file by creating a data directory, downloading the UCI data, and reading it with pandas for initial inspection.

Clean up a 30,000-row, 25-column dataset by preparing explanatory variables and a categorical target—default payment next month—for subsequent payment delay analysis.

Analyze payment delays using a regular expression pattern to identify pay status entries, and visualize the distribution of delays over the past six months with pandas and matplotlib plots.

Explore standing credit by examining past month payments, bill amounts, and pay values, generate bill columns with pattern matching, and compare minimum and maximum credits to illuminate payment behavior.

We explore calculating and plotting payments in the previous months for credit card users. Build pay status columns, limit balance, and use group by with stats and a log-scale axis.

Explore defaulting by visualizing defaulters with a bar graph, group by target, and analyze gender, marriage, and age categories using pd.cut and rename categories.

Explore absolute statistical methods through hands-on visual analysis of a credit card case study, building demographic bar charts by target, sex, marriage, and age category.

Explore feature engineering by creating and transforming dataset columns, applying patterns and distributions, and visualizing bill amount features for a credit card case study.

Create training and testing matrices from 25 attributes for logistic regression, using train_test_split and Patsy formulas, assembling features like sex, marriage, education, age, credit limit, and pay metrics.

Explore visualizations for model evaluation by building confusion matrices and ROC AUC curves alongside logistic and linear regression, with practical plotting workflows.

Learn to create seaborn pair plots to explore correlations in a dataframe by selecting column ranges and setting kind to scatter, using the credit card dataset as an example.