A progressive series of machine learning projects in Python, covering the full supervised and unsupervised learning workflow - from exploratory data analysis through classification, regression, and clustering.
Author: J. Wong
 |
 |
 |
 |
Dataset: Heart Disease - Kaggle (Neurocipher)
Comprehensive EDA on a 270-patient heart disease dataset. Covers data cleaning, correlation analysis, feature engineering, PCA, and frequentist hypothesis testing across four hypotheses.
Topics: Correlation heatmaps · Polynomial & deviation features · PCA (11 components → 95% variance) · t-tests · ANOVA · Chi-square
Dataset: Mobile Price Classification - Kaggle (A. Sharma)
Benchmarks 10+ classification algorithms on a 4-class mobile pricing problem, with full hyperparameter tuning and model interpretability analysis.
Topics: Logistic Regression · KNN · SVM · Decision Trees · Random Forest · Gradient Boosting · XGBoost · LIME · Permutation Importance
Dataset: Student Performance - Kaggle (N. Narayan)
Validates regression assumptions and systematically evaluates regression techniques with cross-validation and regularization. Achieves R² ≈ 0.989.
Topics: Linear & Polynomial Regression · Lasso (L1) · Ridge (L2) · K-Fold Cross-Validation · GridSearchCV · Pipeline
Datasets: Spotify Tracks - Kaggle (MaharshiPandya) · Sign Language MNIST - Kaggle (tecperson)
Applies dimensionality reduction and clustering to both tabular and image data, including pixel-level image segmentation and t-SNE visualization of hand sign embeddings.
Topics: PCA · KernelPCA · t-SNE · KMeans · DBSCAN · MeanShift · Agglomerative Clustering · Gaussian Mixture Models
Methods and workflow inspired by the IBM Machine Learning Professional Certificate on Coursera.
Python scikit-learn pandas NumPy Matplotlib Seaborn Plotly XGBoost SciPy Statsmodels OpenCV KaggleHub