PredictionTableReport

Author	Lily Dong
Creation date	06/01/2011
Firing policy	all
Package	org.seasr.meandre.applet

DESCRIPTION

Provides a visualization to evaluate the performance of a predictive model.Detailed Description: Given a PredictionTable with both the predictions and the actual values, this module will provide a simple visualization to evaluate the performance of a predictive model on a data set. The accuracy, defined as the number of correct predictions, is displayed. A pie chart depicting the accuracy is also shown. A confusion matrix for the PredictionTable is created. The confusion matrix displays the precision and recall of the predictive model. Precision is defined as the number of correct predictions within a class divided by the number of predictions within a class. Recall is the number of relevant predictions within a class divided by the number that actually exist in a class. The confusion matrix also displays the Type I and Type II errors. Type I error is defined as accepting an item as a member of a class when it is actually false, known as a false positive. Type II error is defined as rejecting an item as a member of class when it is actually true, known as a false negative. The confusion matrix is shown with the unique predictions along the top. These are labeled Prediction. The actual values are displayed along the side. These are labeled Ground Truth.Data Type Restrictions: A PredictionTable with both the predictions and actual values is required. This module only supports classification predictions. Continuous predictions are not supported.Data Handling: This module does not modify the input data.Scalability: This module makes one pass over the data to count the number of correct and incorrect predictions.

INPUTS

Name	Description	Example
predictionTable	Read org.seasr.datatypes.datamining.table.PredictionTable with both the actual values and predictions as input.

OUTPUTS
None

PROPERTIES
None