Pyramid Office Hours 4/16: Using advanced analytics in the Discover Module
We had two questions on this topic:
- Is it possible to obtain the model diagnostics? for example, how do we identify which products have been assigned to each cluster?
- Generally in Linear Regression, we use it to make predictions after training with a known dataset
1 reply
-
The advanced analytics functions in the discover module allow you to perform sophisticated analytics in run-time that respond to the context of that report. However, the ability to reuse these models is limited in this context.
We will be adding cluster quality statistics on the a future roadmap update.
If you want to save the model and reuse it on new data you should build it at the model layer. See the link below for documentation on building ML models in the model layer.
The model outputs are available in the discover report allowing you too see how they apply to individual data points.
Categorical models (K-Means Clusters and Outlier Detection) appear under Custom Columns alongside other dimensions making it easy to break out members by clusters.
Numerical Models such as Linear Regressions appear under the measures alongside other custom calculations.
