edX Data Science Professional Certificate

After nine modules covering wrangling, inference, and ML with R, the edX program culminated in two capstone projects that I used to demonstrate modeling depth and documentation rigor.

Project 1 · MovieLens Recommender

• Built a layered recommender system on the MovieLens dataset, progressively adding baseline, user, movie, and temporal effects.
• Tuned regularization terms to keep RMSE low on the validation split; final model achieved 0.8602 RMSE.
• Full report: PDF · R Markdown

Project 2 · Credit Card Fraud Detection

• Tackled a highly imbalanced European transaction set (2013) by pairing SMOTE resampling with cost-sensitive learning to minimize false negatives.
• Explored PCA components (V1–V28) supplied with the dataset to interpret the separation between fraudulent and legitimate behaviour.
• Delivered a narrative report plus the Rmd notebook.
• Final tuned random-forest pipeline caught 100% of known fraud cases with only 33 false positives across 25K transactions.

The write-ups include visual diagnostics (transaction value vs. time, inter-transaction intervals, PCA projections) and detail my approach to stakeholder-friendly reporting.