trendscenter / coinstac-ssr-fsl Goto Github PK

The local stats in coinstac_drne_fsl and coinstac_msr_fsl look similar though

Need to handle this case.

It's also possible in X that the file names are duplicated but the other covariates or y files are not. Or, y is duplicated but the X info is not. I'll filter the unique rows/files and get the first occurrences and then take the intersection of X and y. Or to be safe, just remove all instances of the duplicated rows in both X and y and take it from there.

Especially a docstring for each function.

complete the partially written code in remote.py

https://github.com/MRN-Code/coinstac_ssr_fsl/blob/e57c53189fad92842798460b5030c14c1c3e5138/remote.py#L44

input_list is a dictionary.

To get all dict keys in a list try [site for site in input_list] instead of using list().

https://github.com/MRN-Code/coinstac_ssr_fsl/blob/master/parsers.py#L66

If a file is empty it's corresponding X needs to be removed

Yes at this line- https://github.com/MRN-Code/coinstac_ssr_fsl/blob/master/parsers.py#L19

Write code to handle the aforementioned error:

https://github.com/MRN-Code/coinstac_ssr_fsl/blob/master/parsers.py#L21

Please refer to the following line of code https://github.com/MRN-Code/coinstac_ssr_fsl/blob/master/parsers.py#L25

do we have a default key in the compspec??

Referring to the line of code at https://github.com/MRN-Code/coinstac_ssr_fsl/blob/master/parsers.py#L59 what if y_files is empty? Although this case doesn't even make sense given how the UI is supposed to work, it might make sense to address this case from a code/error-handling standpoint.

The parameter args is not being used in the function. Please remove it.

NaNs in the first column (file names) of X dataframe

Add a docstring and make the name follow PEP 8 (i.e., list_recursive).

Also, I might want to understand the impact of the line X = X * 1.
I really need to work on the parser in great detail.

The problem with doing ridge estimation in this code is that although statsmodels fit_regularized() function can do ridge regression, it doesn't give p-values, t-values, sum-squared-of-errors.

In such a case we need to run a regression ignoring such examples

Talk to Ross about this.

Also refer to the same issue at trendscenter/coinstac-ssr-vbm#3