ML Predictive Modeling Pipeline

The problem

Needed to derive actionable insights from complex datasets while demonstrating both supervised and unsupervised ML competency.

What I built

1. 01

   Performed statistical data exploration, missing-value analysis and visualisations including histograms, box plots, scatter plots and heat maps.

2. 02

   Pre-processed data through imputation, normalisation, standardisation and one-hot encoding.

3. 03

   Trained and evaluated supervised algorithms (linear regression, decision trees, SVMs) using accuracy, precision, recall, F1 and RMSE.

4. 04

   Applied k-means clustering, hierarchical clustering and PCA to identify patterns in unlabeled data.

5. 05

   Improved robustness through k-fold cross-validation and hyperparameter tuning with grid search and Bayesian optimisation.

Project stages