Closed johnbradley closed 1 year ago
The function normalized_df()
removes any columns where all the values are the same:
https://github.com/Imageomics/Andromeda/blob/836608c1cf2925e437e0412d6c08d2c5522f6ba7/andromeda/andromeda.py#L25
This logic came from Andromeda_IMG.ipynb.
Perhaps a good fix would be to remove columns with all the same value from the list of selectable columns. We could add details about why some columns aren't selectable(not numeric or all the same value).
I agree - numeric columns with zero variance should be assumed to be useless for Andromeda.
When clicking "Visualize Data" with datasets/testData/mockData_v1.csv a 500 error is returned. The logs for the flask app show the following error:
If these three fields are unchecked under Configure Visualization the error does not occur. These three columns only contain the value
1
.