Esther Bron - [email protected]
Different types of predictions:
- for the final TADPOLE dataset (a,b) or for the leaderboard dataset. (c,d)
- based on multiple timepoints (longitudinal data) (a,c) or one timepoint (crossectional data) (b,d)
- script_dataPrep.py with setting: leaderboard = 0
- TADPOLE_SVR.py with settings: leaderboard = 0, d3 = 0
Inputs: TADPOLE_D1_D2.csv, D4_dummy.csv (Fake reference standard to test validity of output), FeatureIndices.csv (Result of feature selection performed in R) Intermediates: LongTADPOLE.csv (Training and Test data), ToPredict_D2.csv (IDs van test set) Outputs: TADPOLE_Submission_EMC-EB1.csv
- script_dataPrep_D3.py
- TADPOLE_SVR.py with settings: leaderboard = 0, d3 = 1
Inputs: TADPOLE_D1_D2.csv, TADPOLE_D3.csv Intermediates: LongTADPOLE_D3.csv (Test set), LongTADPOLE_D1.csv (Training set) Outputs: TADPOLE_Submission_EMC-D3-EB1.csv
- script_dataPrep.py with setting: leaderboard = 1
- TADPOLE_SVR.py with settings: leaderboard = 1, d3 = 0
Inputs: TADPOLE_D1_D2.csv, TADPOLE_LB1_LB2.csv, TADPOLE_LB4.csv (Reference standard test set), FeatureIndices.csv (Result of feature selection performed in R) Intermediates: LongTADPOLE.csv (Training and Test data), ToPredict.csv (IDs van test set) Outputs: TADPOLE_Submission_Leaderboard_EMC-EB1.csv
Not implemented (leaderboard = 1, d3 = 1).
[[111 120 75] [135 106 71] [ 96 106 76]] Diagnosis: mAUC = 0.503 BAC = 0.496 ADAS: MAE = 5.368 WES = 4.692 CPA = 0.492 VENTS: MAE = 8.584e-03 WES = 8.283e-03 CPA = 0.499
[[ 72 186 48] [ 78 198 36] [ 62 170 46]] Diagnosis: mAUC = 0.513 BAC = 0.520 ADAS: MAE = 0.947 WES = 0.921 CPA = 0.473 VENTS: MAE = 9.545e-04 WES = 9.616e-04 CPA = 0.482
[[194 1 0] [ 54 96 0] [ 6 49 17]] Diagnosis: mAUC = 0.840 BAC = 0.901 ADAS: MAE = 4.541 WES = 4.168 CPA = 0.490 VENTS: MAE = 3.513e-03 WES = 3.304e-03 CPA = 0.494