This document summarizes major improvements and fixes introduced in the Accelerate Win Rate Optimization package release version.
|
Version |
1.0 |
|---|---|
|
Release Date |
New Features and Improvements
|
New Feature Description |
ID |
|---|---|
|
A heatmap chart is now available to show the Pairwise impacts of features on the Win Rate model. |
PFPCS-9416 |
|
A portlet in the Definition tab now summarizes data rows filtered out due to invalid Win/Lost values. |
PFPCS-9449 |
|
The dropdown menus for Win/Lost values are now limited to the first 100 distinct values to prevent the UI from freezing. |
PFPCS-9476 |
|
If discount % is selected without Revenue at List Price, the model informs the user and removes discount % from the model. |
PFPCS-9489 |
|
The accelerator now checks for and forbids the use of reserved names or selecting the same feature as both categorical and numerical. |
PFPCS-9490 |
|
Expired feature tables from previous calculations are now automatically dropped before a new training run. |
PFPCS-9512 |
|
The Pairwise Impact inputs now dynamically allow only existing pairs while keeping full choices for Feature 1 for better UX. |
PFPCS-9516 |
|
The model evaluation now displays the optimized price. |
PFPCS-9528 |
|
The Evaluation Dashboard now includes a feature importance view to highlight win rate drivers. |
PFPCS-9529 |
|
The feature importance visualization now uses a red-to-green color scale. |
PFPCS-9536 |
|
Constant features are handled by removing them from the model and informing the user. |
PFPCS-9545 |
|
Using a cost of zero is forbidden to avoid invalid profit calculations. |
PFPCS-9601 |
|
The evaluation allows entering and assessing a new categorical value. |
PFPCS-9695 |
|
Price optimization supports targeting either revenue or profit. |
PFPCS-9824 |
|
Numerical inputs below the training minimum are handled using the minimum’s adjustment in Evaluation. |
PFPCS-9837 |
|
The price adjustment logic correctly handles high price values. |
PFPCS-9898 |
|
Revenue and profit are split into separate charts in Evaluation. |
PFPCS-9899 |
|
All chart portlets now include a Data tab. |
PFPCS-9960 |
Fixed Issues
|
Bug Description |
ID |
|---|---|
|
The density bar for null or empty string values in categorical features displays incorrectly in the Feature Impact chart. |
PFPCS-9683 |
|
An "Invalid input value" error for an unknown category occurs in the Evaluation tab. |
PFPCS-9889 |
|
The Python script fails when the test set contains only wins or only losses. |
PFPCS-9896 |
|
The win rate curve in the "Win Rate and Optimal Price" portlet incorrectly displays values at its extremities. |
PFPCS-9929 |
|
Creating temporary tables with long feature names causes an error. |
PFPCS-9992 |