causality

Report for Gavin Cawley

Download Fact sheet

 Performance as a function of the number of features

Figure 1: Performance as a function of the number of features. (Click to enlarge). Legend: red: REGED, green: SIDO, blue: CINA, black: MARTI. circle: version 1 (unmanipulated test set), full circle: version 2 (manipulated), star: version 3 (manipulated)

 Profile of features selected

Figure 2: Profile of features selected. (Click to enlarge). Legend: dcause=direct cause, deffect=direct effects, ocauses=other causes (indirect), oeffects=other effects (indirect), spouses=parent of a direct effect, orelatives=other relatives, unrelated=completely irrelevant

 Tscore performance of all entries

Figure 3: Tscore of all entries made as a function of entry order. (Click to enlarge). The dashed lines indicate the 25 and 50 percentiles.

 Fscore performance of all entries

Figure 4: Fscore of all entries made as a function of entry order. (Click to enlarge). The dashed lines indicate the 25 and 50 percentiles.

Performances:

Fnum=number of features (*=sorted list, **=table of results. Fnum given for best Tscore if table of results submitted), Fscore=MB AUC, Tscore=AUC of test predictions; Top Ts=best score among all valid last entries made by participants. Max Ts=best score reachable, including reference entries using the knowledge of true causal relationships not available to participants.
* Entry Method Fnum Fscore Tscore (Ts) Top Ts Max Ts <Tscore> Rank
REGED0 1373 final models 128/999 ** 0.9410 0.9997±0.0012 1.0000 1.0000
REGED1 1373 final models 32/999 ** 0.8393 0.9787±0.0036 0.9980 0.9980 0.9276 2
REGED2 1373 final models 8/999 ** 0.9985 0.8045±0.0056 0.8600 0.9534
SIDO0 1373 final models 4928/4932 ** 0.5890 0.9427±0.0070 0.9443 0.9467
SIDO1 1373 final models 4928/4932 ** 0.5314 0.7532±0.0137 0.7532 0.7893 0.7881 1
SIDO2 1373 final models 4928/4932 ** 0.5314 0.6684±0.0130 0.6684 0.7674
CINA0 1373 final models 128/132 ** 0.5166 0.9743±0.0031 0.9788 0.9788
CINA1 1373 final models 128/132 ** 0.5860 0.8691±0.0046 0.8977 0.8977 0.8488 3
CINA2 1373 final models 64/132 ** 0.5860 0.7031±0.0047 0.8157 0.8910
MARTI0 1373 final models 128/1024 ** 0.8697 0.9996±0.0012 0.9996 0.9996
MARTI1 1373 final models 32/1024 ** 0.8064 0.9470±0.0039 0.9470 0.9542 0.9147 1
MARTI2 1373 final models 64/1024 ** 0.9956 0.7975±0.0059 0.7975 0.8273

Return to Causation and Predition Challenge Analysis