--bootstrapp

Switch

--bootstrapp

Description

Currently (2015-07-24) only implemented for --AUC switch of --correlate

Argument and Default Value

Non-parametric significance test. On each iteration, shuffles predictor variable relative to outcome (and, if given, outcome controls) to get null distribution of AUC values. p-value is reported as percentile of actual value relative to the null distribution (i.e., percentage of values higher than real value)

Details

Other Switches

Required Switches:

• --correlate

Optional Switches:

• --no_correction

• --p_correction

• --outcome_controls

Example Commands

dlatkInterface.py -d tester7 -t statuses_er1 -c study_code --group_freq_thresh 500 \
-f 'feat$cat_ser1_f2_200_cp_w$statuses_er1_655$study_code$16to16' --outcome_table outcomesFinal \
--outcomes DM_UNCOMP --correlate --rmatrix \
--controls sex_int isWhite isBlack isHispanic ageTercile0 ageTercile1 ageTercile2 \
--AUC --bootstrapp 10000 --no_correction --csv --output_name OUTPUT