Home
psptools Reference Guide
The Tandy Center for Ocean Forecasting presents an R package for predicting shellfish toxicity by grouping together consecutive samples at unique monitoring stations.
Setup First time users should start here to configure their environment
Explore the Input Data Take a look at the data set used as an example in this walk through
Build Training Samples Unpack the algorithm that builds a set of training samples
Prepare Model Input Prepare the model input the way that Tensorflow expects
Prediction Fit a model and test it on a subset withheld from the training set
Manage Model Configurations Track of all the parameters of our input data and model architecture
Explore Class Imbalance How imbalanced is our training set and are there ways to deal with it?
References Helpful links to project related materials
PSPverse - family of packages
pspdata - Data pipeline for forecast system with connection to database and formatting functions
psptools - Tools for transforming raw monitoring data into model ready input; building, refining and loading models; running configuration experiments (hindcasts) and making live predictions in forecast mode
pspforecast - Serves and formats predictions to end-user delivery platforms