Release Summary - January 2020
When creating strategies with Data Science Workbench (DSW), and you have a significant percentage of products for the model that are non-recommendable, then your strategy will not be published. This was changed as of the first January release, where you can now adjust the threshold for recommendable products for your DSW strategies on the model builds page. For example, if you set the threshold to 95% of products are recommendable, 5% of the products can non-recommendable and still publish the model.
The PurchaseCPV2 model has been added to the model builds page, where you can enable the model and the strategies dependent on the model for your site. The PurchaseCPV2 model applies a dampening factor to top selling products to reduce the occurrence of products in the model. This model is very useful for grocery sites for example where certain products are purchased quite often with most of the products. The model prevents these top sellers from being heavily recommended.
The PurchaseCPOmniModel and PurchaseCPOfflineModel have been added to the model builds page enabling you to configure the minimum number of co-purchase events for the model. The default value is 1 co-purchase event in the last 100 days. Raising the number will reduce the number of products recommended, but will ensure that the strategy recommends products that are purchased together more often during the last 100 days.
The BestOfferModel has been added to the model builds page enabling you to have strategies with only the products that match certain attributes, attribute values, or condition identified. For example, you can configure the model to include all products that are on sale or all products that have specific attribute or attribute value, such as being part of a promotion. Products are sorted by top sellers.
Release summaries have been added to the home page of the dashboard. The release summary summarizes the new features or major issues fixed in releases as of that month. Release summaries cover releases as of the 1st week of the month.