Skip to content

2'''. Generating the predictive model

Romain F. Laine edited this page Nov 25, 2018 · 1 revision

Performed by: ML_GenerateModel_RF.m

This code generates the random forest model using the training (manually annotated) dataset.

Input: This requires the output from manual curation step (LabelObjects_BATCH.m), which can be pruned (PruneLearners.m) and augmented (CombineTrainingDataset_Duplicate_BATCH.m).

Paramaters: The list of descriptors to use by this model should be set (the pruned descriptors should be selected here). The number of fold for cross-validation, the number of sample for posterior probability calculation, the number of splits and epochs can be set.

Output: The model is saved as a .mat file. The confusion matrix can be saved as well.