PORTFOLIO

FEATURED PROJECTS

01 — STOCK PRICE PREDICTION

ARIMA vs LSTM

Built and compared a statistical ARIMA model against a deep sequential LSTM network for stock price movement and volatility forecasting.

Model Performance LSTM ACHIEVED A LOWER RMSE
02 — SENTIMENT ANALYSIS

Understanding the Voice of Twitter

Logistic Regression on a public Kaggle Twitter dataset using TF-IDF (unigrams + bigrams, 5,000 features).

Accuracy Metric ~70% TEST ACCURACY
03 — GAME RECOMMENDATION ENGINE

Semantic Similarity with BERT

Content-based recommender trained on game metadata. BERT embeddings generated per game, cosine similarity used for recommendation ranking.

User Engagement ~60% HIT RATIO