Problem
The rolling file currently uses a fixed number of records when loaded during the readFromBin method on the DriftModel class, this fixed number may not be a flexible enough method of selecting the number of records to compare for rolling drift. It would be an improvement to modify this so that it uses entries from the last N days instead.
Solution
Building this would require adding a date to the binary file write method to store when a file was added, or replacing the binary file writer entirely and switching to a different file storage format. Regardless, at the time of this readme writeup, there is no date associated with stored vectors and so there is no way to have the rolling file use the last N days of records instead of a fixed count.
Additional information
Its possible this could be a setting that can be toggled using a global boolean variable, but it becomes a little sticky when the variable needs to update both python and javascript file loading.
👨👧👦 Contributing
Problem
The rolling file currently uses a fixed number of records when loaded during the readFromBin method on the DriftModel class, this fixed number may not be a flexible enough method of selecting the number of records to compare for rolling drift. It would be an improvement to modify this so that it uses entries from the last N days instead.
Solution
Building this would require adding a date to the binary file write method to store when a file was added, or replacing the binary file writer entirely and switching to a different file storage format. Regardless, at the time of this readme writeup, there is no date associated with stored vectors and so there is no way to have the rolling file use the last N days of records instead of a fixed count.
Additional information
Its possible this could be a setting that can be toggled using a global boolean variable, but it becomes a little sticky when the variable needs to update both python and javascript file loading.
👨👧👦 Contributing