Explore Data Insights

You can explore insights for any model to which you have access. An insight is a statistically significant finding in your data. When you create a model version, Einstein Discovery analyzes the data in your dataset and generates insights based on its analysis. Insights provide a starting point for you to investigate the relationships among your model’s explanatory variables and its goal.

• Navigate Data Insights
  Learn about the data insights interface and how to interact with it.
• Explore Insights for a Variable
  Explore how an explanatory variable is related to your model’s goal.
• Compare Subgroups with the Global Average
  Explore how a subgroup (category or bucket) relates to a variable’s global average. Isolating a subgroup helps you better understand how it relates to the model’s goal.
• Explore Why a Value Does Better or Worse Than Average
  Learn why a category or bucket does better or worse than the global average.
• Compare Subgroups
  Compare subgroups (categories or buckets) to learn how they relate to your model’s goal. For example, you can compare the sales performance between two different territories (such as “North America” and “Europe”). Einstein Discovery displays a waterfall chart to help you visualize the comparison between territories. Comparative insights help you determine whether the patterns you focus on are real and not accidents of the data.
• Explore Predictions and Improvements
  Explore predicted outcomes and suggested ways to improve those predicted outcomes. Perform interactive, “what if” analyses and change feature selections to see prediction scores, top prediction factors, and top improvements to enhance prediction scores.
• Explore Insights for Multiclass Outcomes
  Learn what Einstein Discovery insights can tell you about your multiclass classification solution.
• Explore Bias in Multiclass Outcomes
  Bias detection in multiclass models alerts you of the presence of bias so you can mitigate its distorting effects in your analysis.
• Explore Predictions for Multiclass Outcomes
  Explore predicted outcomes for multiclass models. Perform interactive, “what if” analyses and change feature selections to see the winning prediction, ranked list of outcomes by probability, model accuracy, and top prediction factors.
• Bookmark an Insight in a Model
  Bookmarking an insight allows you to export it to a Quip document. It also makes it easier to quickly find it in the Insights List when the Bookmark filter on the toolbar is disabled.
• Get Descriptive and Diagnostic Insights Programmatically
  Use Einstein Discovery's Insights API to embed insights from your model into any website, application, or dashboard. Insights contain values that quantify the impact on the selected model goal, along with narrative descriptions of those values. With the Insights API, client applications can obtain a summary of the top positive and negative insights, query insight details further, and then apply their own narratives to the insight values. You can also query insights by insight type (descriptive or diagnostic).