Forecasting Methods

Catalog description

Forecasting Methods will provide the engineering student with the skills necessary to perform data driven time series analysis from an engineering applications perspective. A process driven approach will be used covering the entire forecasting process from data preparation and pre-processing techniques to model selection, performance evaluation, and monitoring. A special emphasis will be placed on performance evaluation and improvement of models used to predict RIT energy demand and peak load days. The course will cover topics in data cleansing, data transformation, trend and seasonality analysis, smoothing techniques, regression analysis for forecasting, seasonal and non-seasonal ARIMA models, dynamic regression, neural networks and advanced modeling techniques for multivariate time series analysis. Lectures and assignments will focus on predicting RIT energy demand considering circuits with 2MW solar fields or similar data sets.