An end-to-end audio ML pipeline testing whether a classifier trained only on Mel-Frequency Cepstral Coefficients (MFCCs) can detect code-switching across three social contexts (family, close friend, professional) from personal WhatsApp voice notes. Compares a from-scratch Multinomial Softmax Regression model against a 1D CNN (Keras) and evaluates with confusion matrices, ROC curves, and per-class F1 scores.
"}},{"node":{"frontmatter":{"title":"Hierarchical Modeling for Sports Attendance","tech":["Python","Jupyter Notebook","Bayesian Hierarchical Modeling","Negative Binomial Regression","WAIC / LOO Cross-Validation"],"github":"https://github.com/Shazil10/Discrete-and-Multilevel-Models","external":"/cs146-discrete-multilevel-models.pdf"},"html":"A Bayesian count-modeling project that analyzes attendance for 12 professional sports teams (218 observed games) and predicts 22 missing attendance values from ticket-scanner failures. Compares a complete-pooling baseline against a hierarchical model with team and day-of-week effects using a Negative Binomial likelihood for overdispersion, and selects the better specification via WAIC and LOO cross-validation before generating posterior predictive estimates for the missing games.
"}},{"node":{"frontmatter":{"title":"Bayesian Linear Regression for Taipei Housing Prices","tech":["Bayesian Linear Regression","Model Comparison","Cross-Validation","Housing Price Modeling"],"github":"https://github.com/Shazil10/CS146-Linear-Regression-Assignment","external":"/cs146-linear-regression.pdf"},"html":"A Bayesian linear regression study of 414 housing transactions in New Taipei City's Sindian District, comparing multiple models of how MRT distance, building age, and local amenities affect unit prices. Uses cross-validation to evaluate competing specifications (simple vs flexible distance effects and neighborhood terms) and quantifies uncertainty in effect sizes for real-estate decision making.
"}},{"node":{"frontmatter":{"title":"CoreWeave Equity Valuation (DCF, Comps & Precedents)","tech":["DCF Valuation","WACC & Terminal Value","EV/Revenue Multiples","Precedent Transactions","Sensitivity Analysis","Excel Modeling"],"github":"https://github.com/Shazil10/CoreWeave-Growth-Valuation","external":"/coreweave-growth-valuation.pdf"},"html":"An equity valuation of CoreWeave (CRWV) following its March 2025 IPO, triangulating intrinsic value using a DCF (with WACC/terminal growth sensitivities), public comparables (EV/Revenue), and precedent transactions. Highlights concentration and leverage risks and synthesizes results into a football-field range and investor recommendation.
"}},{"node":{"frontmatter":{"title":"Berlin Traffic Congestion Modeling","tech":["Python","OSMnx","NetworkX","Simulation","Data Visualization","Congestion Modeling"],"github":"https://github.com/Shazil10/Berlin-Traffic-Congestion-Simulation","external":"/berlin-traffic-congestion-simulation.pdf"},"html":"An applied traffic simulation on the Berlin road network using OSMnx and NetworkX shortest-path routing with edge travel-time weights. Compares threshold-based “jam” rules with gradual and density-based congestion slowdowns, and builds enhanced visualizations (hotspots, flow arrows, legends, and animation) to diagnose network bottlenecks over time.
"}},{"node":{"frontmatter":{"title":"Bunge Global SA: Capital Budgeting & FX Risk Strategy","tech":["Capital Budgeting (NPV/IRR)","FCFF vs FCFE","WACC / Cost of Capital","FX Scenario Analysis","Hedging Strategy Design","Sensitivity Analysis"],"github":"https://github.com/Shazil10/Bunge-Global-Financial-Strategy","external":"/bunge-global-financial-strategy.pdf"},"html":"A global financial strategy case study on Bunge Global SA (BG) proposing and valuing two investments: a 20,000-ton pea protein facility and a digital agribusiness platform. Builds FCFF/FCFE models to evaluate NPV/IRR/payback, stress-tests outcomes across operational and financing assumptions, and models multi-currency revenue exposure with FX scenarios and hedging strategies to preserve risk-adjusted returns.
"}},{"node":{"frontmatter":{"title":"Bacteria Growth and Diffusion Simulation","tech":["Python","NumPy","Matplotlib","Grid Simulation","Diffusion Models"],"github":"https://github.com/Shazil10/Bacteria-Growth-Model","external":"/bacteria-growth-model.pdf"},"html":"An implementation and analysis of a 2D grid-based bacteria growth model coupled with a renewable food resource. The system applies logistic food regrowth with diffusion, bacteria consumption with starvation dynamics, reproduction, and bacteria diffusion, producing emergent spatial patterns and validated via targeted test cases.
"}},{"node":{"frontmatter":{"title":"Airport Security Queue Simulation","tech":["Python","NumPy","Pandas","Matplotlib","Discrete-Event Simulation"],"github":"https://github.com/Shazil10/Airport-Security-Queues","external":"/airport-security-queues.pdf"},"html":"A discrete-event simulation of an airport security checkpoint operating over 24 hours, where travelers arrive via a Poisson process and route to the shortest queue. Models a senior officer bottleneck (3% of travelers requiring additional screening) and validates empirical queue lengths against M/G/1 theory using the Pollaczek–Khinchine formula.
"}},{"node":{"frontmatter":{"title":"Gamma Modeling of Football Waiting Times","tech":["Probability & Statistics","Exponential Distribution","Gamma Distribution","Goodness-of-Fit Diagnostics"],"github":"https://github.com/Shazil10/CS114-Randomness-Boca-Juniors","external":"/cs114-randomness-project.pdf"},"html":"An empirical study of the inter-arrival times between ball-out-of-play events in Boca Juniors Feminine matches, treating them as realizations of a random process. Uses descriptive statistics, histogram/PDF fits, CDF comparisons, Q–Q plots, and boxplots to compare Exponential vs Gamma models, concluding that evolving match conditions produce waiting times best captured by a Gamma distribution rather than a simple memoryless process.
"}}]}}}