search

thekingofkings / chicago-crime

Crime correlation anaysis
MIT License
11 stars 3 forks source link

The feature significance is not consistent comparing NB with GWNBR #20

Open thekingofkings opened 7 years ago

thekingofkings commented 7 years ago

First re-run previous NB permutation test, and check the results. TODO: implement GWNBR and test the significance.

thekingofkings commented 7 years ago

Issue #16 is actually a prerequisite for this task. Need finish that one first.

thekingofkings commented 7 years ago

2000 iterations of permutation. The p-values are:

{'taxi': 0.0525, 'demo': 0.0, 'geo': 0.0155, 'poi': 0.0}

thekingofkings commented 7 years ago

The permutation results are from NB model. Need to test GWNBR as well.

thekingofkings commented 7 years ago

Use GWNBR to test the significance.