Version 2.0 (September 2023)
Consult the SpP project documentation here.
Version 2.1 (November 2023)
Here is a detailed description of all the changes in version 2.1, which will be released in November:
Code-base:
- Addition of a new function for the market segment feature creation : [PO-1311]
- Move of the target_computation class to code-base : [PO-1320]
- Addition of a new function for values isolation from a field using project variables : [PO-1366]
- Integration of a new function of the code base to be able to bin numerical features : [PO-1387]
Dataiku project
The new “Manual_regions” dataset is first integrated from the compute_Price_drivers_features recipe, in which the update_manual_inputs() function is used to update the data with the new entries. Then, the synchronization recipe is applied to refresh the file with “Manual_regions_output”.
- New flow zone for Data drift analysis : [PO-1319]
that contains a new python recipe for preparing reference data, in which we can specify the number of versions to extract from the Features folder (using the output All_run_versions_dataset_output), and a standalone evaluation recipe to perform data drift analysis.
- Split of Galden into its own family and properly update the variables to take in account the new family : [PO-1383]
- Refine algorithm by incorporating the feedback that was collected during our conversations with SpP PMs : [SpP] Refine core algorithm