I’d like to suggest an enhancement for the Constant Value Column Filter. Currently, it only removes columns with a single value across all rows. It would be great to have a feature similar to the Missing Value Column Filter, where we can set a threshold for the percentage of constant values to trigger column deletion.
This would help manage columns with repeated values that aren’t uniform but still unnecessary, simplifying our workflows without needing to convert these values to missing before filtering.
I applied the low variance filter to my data frame, but some columns still remain. In my case, certain batches (rows) have mostly zeros in specific columns, while others show varying values. These zero-heavy rows invalidate the measurements in those columns. Even if other rows are fine, a column should be eliminated if most values in a specific batch are zero. However, since I have many batches, one batch being mostly zero doesn’t significantly affect the variance in that column, making the low variance filter ineffective.
To address this, I converted zeros to missing values and used the missing value column filter with a minimum threshold. It would be easier if the Constant Values Column Filter node had an option to filter columns based on a threshold of repeated values, allowing for more effective column depletion.