Prophesy Summer Release Announcement (Version 3.2)
After several weeks of intense development and testing, Prophesy Version 3.2 was released on July 7, 2017. This release implements over twenty product improvements based on your feedback. Here are some of the highlights:
- More user control over data feeds. For example, you can attach names and detailed descriptions to data feeds, remove unwanted feeds, and cancel feeds without waiting for them to complete.
- Pre-validation of data feeds: When a dataset is submitted, Prophesy performs over a two dozen validation checks and reports any issues immediately.
- Multiple alternatives to exclude variable outliers.
- Selection of correlated variables. Prophesy automatically recognizes groups of highly correlated variables, selects the one with best fit and excludes the others. You have the option to select a different variable from a group if necessary.
- New algorithms added to the model catalog. We added some advanced methods which have recently emerged from the research community, further improving the accuracy of our best predictive models by 5 to 10%.
We also made many performance and UI improvements. We invite you to take advantage of these new features, and welcome your feedback – your suggestions could become new features in our next version!