Improve Performance of AUC better than this
Added Clip values – This removes the outliers where you can specify both lower and upper boundaries. Values which are outside the range are removed. In this experiment clip values are used to clip peaks and subpeaks with percentile value to constrain the column values t percentile range where Only AA4, AA7, AA14, AA15 columns were included. Used edit Metadata to make changed the data type to integers so that column metadata remains unchanged and changed it to categorical so that columns will be treated as categories Split Data was changed – unchecked the randomized split to avoid the rows to be randomly assigned to folds and changed the random seed to 3456 so that the rows could be divided the same way every time. Two decision jungle gives the better performance. Two decision jungles: changed the resampling method to bagging and create trainer mode was changed to single parameter to provide specific set of values to arguments. Changed the number of DAGs to 70 where min width was set to 250 to max width-20148 Added Permutation feature importance which is used to score and evaluate features. Used Classification metrics to evaluate accuracy, Precision, Recall, Average Log Loss.