d909b / perfect_match Goto Github PK

This is a fascinating work! Can this method be extended to the scenario where the treatment is continuous, say, the degree of medication for a patient? One easy modification is to discretize the continuous variable.

Hi,
today I tried to play around with this tool with the ihdp data set, but it shows this kind of error when i run the data.

This is the error looks like:

...
    question_marks=",".join(["?"] * len(values[0]))), values)
sqlite3.InterfaceError: Error binding parameter 1 - probably unsupported type.

Maybe the code data_access/ihdp/data_access.py at line 128, x not to be defined as ARRAY?

Hello, I am doing some expreiments about causal inference. And I find jobs dataset in /perfect_match/data_access/jobs/, so I use them jobs_DW_bin.test.npz and jobs_DW_bin.train.npz. But I find there 17 covariates in your dataset, while ony 8 covarites in raw data(Lalonde dataset). May I ask why.
Thanks a lot!
Qidong Liu