Error analysis dashboard
Error analysis allows us to analyze where the model is underperforming and also find errors in the decision-making process.
Error analysis can be done as either a tree map or a heat map. Based on the analysis, red represents an error. If you click one of the insight bubbles, you’ll see the path the model traversed.
Dashboard analysis will be based on the type of ML modeling we choose. The two types are classification and regression. In our example, we chose regression. The following are two of the accuracy metrics used:
- Mean squared error
- Mean absolute error
The following screenshot shows the error analysis dashboard UI created by the Responsible AI SDK.
Figure 8.12 – Error analysis dashboard
This dashboard provides options to save the error analysis for further analysis or share it with others. This feature is very helpful if we need to share information with other data scientists or subject...
