USGS Workshop Video Training

Workshops led by Wisconsin Sea Grant and the USGS Wisconsin Water Science Center titled Developing Predictive Models for your Beach: Hands-on Training, were recorded and are available below. The full training for building a complete model in Virtual Beach takes 10-12 hours.

Introduction to Predictive Modeling  (0:55:01)
Basic concepts and methods of modeling beach water quality.

Model Uncertainty and Stability  (0:22:19)
Overview of the major sources of model uncertainty.

Data Processing-Part 1 (0:51:19)
A. Determine the orientation of your beach
B. Import and clean up model-building data for your beach

Data Processing-Part 2 (0:34:04)
C. Process directional data (wind, currents, waves)
D. Combine two or more predictive variables
E. Transform variables and explore potential relationships 
F. Save a Virtual Beach project file

Building & Evaluating ‘Anytime’ Nowcast Models  (1:09:09) 
A. Open a previously saved VB project file 
B. Build and evaluate an anytime GBM model 
C. Optimize a GBM model by removing variables 
D. View a GBM model within the VB Prediction tab 
E. Save a GBM model into a new VB project file

Building & Evaluating Sampling-Day Nowcast Models-Part 1  (1:09:32)
A. Build and optimize an MLR model 

Building & Evaluating Sampling-Day Nowcast Models-Part 2  (1:04:25)
B. Evaluate top-rated MLR models
C. Set MLR decision criteria
D. Evaluate MLR residuals and search for influential outliers

Nowcast Prediction & Validation-Part 1  (0:32:54)
A. Create a real-time Web data services (EnDDaT) connection 
B. Download real-time data to make a nowcast prediction

Nowcast Prediction & Validation-Part 2  (0:30:50)
C. Import a historical data table to validate your models
D. Save a VB model file with prediction results