Python for Machine Learning & Data Science Masterclass

Learn Python basics: numbers, types, and basic operations, plus strings, prints, and string formatting. Explore lists, indexing, dictionaries, booleans, tuples, and sets, with emphasis on data structures and common pitfalls.

Master Python basics by exploring comparison and logic operators, if/elif/else control flow, for and while loops, range usage, and list comprehension with emphasis on indentation.

Learn to create NumPy arrays from Python lists and built-in functions, and generate random data with NumPy. Explore array attributes, shapes, and methods like reshape, max, min, argmax, and argmin.

Explore a tips data frame in pandas, inspecting columns, index, and basic attributes with df.head, df.tail, and df.info; compute descriptive statistics using df.describe and experiment with transpose for readability.

Explore groupby operations in pandas to aggregate data per category, using model year and cylinders, and work with multi-level indexes and cross sections (xs).

Explore combining dataframes with pandas pd.concat, across columns or rows, handling missing values, and preparing for future merge concepts.

Master left and right merges in pandas by joining registrations and logins on name, showing how left preserves all left rows with null values, while right preserves all right rows.

Explore pandas dt time methods to extract year, month, day, and day of week, enabling feature engineering, date parsing with to_datetime, and time-based resampling.

Explore pandas project exercise solutions: read data into a hotels dataframe, compute missing data, drop columns, use value_counts, idxmax, and groupby; derive total spend and arrival day counts.

Create a blank Matplotlib figure canvas, add axes, and plot data, then adjust x and y limits, labels, and titles for multiple plots.

Explore seaborn grids to automatically generate subplot grids from categorical data using catplot and PairGrid, map plots to grid regions, and customize with hue and legends.

Define why we use machine learning and distinguish supervised from unsupervised learning. Explore the supervised learning process and key topics like performance metrics, cross-validation, and feature engineering.

Learn how machine learning uses data-driven algorithms to infer features and labels with estimators that predict outcomes like credit scores and house prices.

Identify the two main machine learning types: supervised learning uses historical labeled data to predict future values, while unsupervised learning uses unlabeled data to discover patterns.

Apply scikit-learn linear regression to relate TV, radio, and newspaper spend to sales. Learn data preparation, train-test split, model training, and interpreting beta coefficients to compare advertising channels.

Assess linear regression validity by visualizing residuals against true Y values—via residual plots and distributions—to detect nonlinearity or issues in datasets like Anscombe's quartet.

Save, load, and deploy a linear regression model with joblib; interpret tv, radio, and newspaper coefficients and forecast campaigns. Evaluate mean absolute error and root mean squared error.

Train and evaluate polynomial regression by transforming three features into nine via polynomial features, then compare MAE and RMSE with linear regression. Save the polynomial converter for future data points.

Explore the bias-variance trade-off in supervised learning, from overfitting and underfitting to understanding how model complexity and training versus test error guide polynomial regression and other algorithms.

Learn how regularization reduces overfitting by adding bias and penalties to the loss, and compare L1, L2, and elastic net with lambda, alpha, and cross-validation to tune them.

Explore ridge regression as a regularization technique for linear regression, adding a shrinkage penalty on beta coefficients to reduce overfitting, with lambda tuned via cross-validation.

Explore L1 regularization with Lasso regression, learn how LassoCV selects alpha values via cross-validation, and compare sparsity and interpretability with Ridge in Python.