Statistical analysis comparing six regression models (Multiple Linear Regression, Ridge, PLS, SVM, Random Forest, and Gradient Boosting) to predict pH levels in a beverage manufacturing process. Random Forest was selected as the final model based on validation performance (RMSE = 0.0995, R² = 0.649). The analysis identifies Manufacturing Flow Rate, Usage Rate, Bowl Setpoint, Filler Level, and Temperature as the top predictive factors. Created for regulatory compliance and process optimization.

Analysis of tree-based modeling techniques including random forests, gradient boosting, conditional inference trees, and Cubist using simulated data and chemical manufacturing process data. Exercises from Kuhn & Johnson's Applied Predictive Modeling.

Nonlinear regression modeling (Neural Networks, MARS, SVM, KNN) applied to Friedman simulation data and chemical manufacturing yield optimization. Demonstrates MARS's automatic feature selection and compares nonlinear vs. linear model performance for pharmaceutical applications.

Applied predictive modeling assignment: Linear regression methods (PLS, PCR, Ridge, Lasso, Elastic Net) applied to pharmaceutical permeability prediction and chemical manufacturing yield optimization.

This project applies time series forecasting techniques to two real-world datasets: ATM cash withdrawals and residential electricity consumption. Part A forecasts daily cash withdrawals for four ATM machines in May 2010 using ARIMA and ETS models, with weekly seasonality patterns identified and incorporated. Part B models monthly residential power usage from 1998-2013 and generates 2014 forecasts, featuring seasonal decomposition and comparison of ARIMA, ETS, STLF, and seasonal naive methods to capture strong monthly patterns driven by heating and cooling demands. The analysis includes data cleaning, exploratory visualization, model comparison, residual diagnostics, and forecast validation.

Exploring ARIMA Models

Exploring Exponential Smoothing using Hyndman's Forecasting Principles Textbook Exercises

Analysis of benchmark forecasting methods using R and the fpp3 package.

Forecasting: Principles and Practice (3rd ed): Chapter 3: Time Series Decomposition Exercises, from Hyndman textbook.

Predictive Analytics Exploration of Hyndman's Forecasting Principles and Practices textbook Chapter 2 Exercises