PredictionResultsViewer (v5) BETA

Summary:

The PredictionResultsViewer is used to view results from testing a predictor where the results can come from either a separate test set (e.g. the KNN or WeightedVoting algorithm) or a cross-validation prediction (e.g., the KNNXValidation or WeightedVotingXValidation algorithms). The prediction results and the model used to obtain them are summarized in the top portion of the panel. Details of the prediction results for each of the samples are shown in the table at the bottom of this panel.

The first column in the table gives the sample name (datapoint). The second column gives the true class from the class template file and the third column gives the predicted class. The fourth column gives the Confidence associated with the predicted class (see algorithm documentation for a description of its calculation.) The fifth column indicates when the prediction is an error or no call is made.

The confidence threshold for making a call and potential errors can be adjusted by entering a number into the No Call Threshold box and pressing the Update button. The confusion matrix displays a table that summarizes the predicted classes against the true classes. 

The viewer includes an interactive plot of the confidence in the prediction results. Mousing over a point in the plot displays the sample name and confidence value, as well as the color associated with its predicted class. Users can zoom in and out of the plot by clicking and dragging the mouse. The scale of the plot can be reset to its original value by double-clicking in the plot. 

References

[appropriate papers cited]

Known Issues/Coming Features:

*for internal use

Parameters

* - required

Input Files

1. Prediction Results File
   Prediction results file in .odf format.​

Example Data

https://software.broadinstitute.org/cancer/software/genepattern/data/protocols/all_aml_test.cart.pred.odf

Requirements

JavaScript

Task Type:
javascript

CPU Type:
any

Operating System:
any

Language:
JavaScript