Visualization
You can use Pyreal's `visualize` module to quickly get explanation graphs
The full code for this and all other user guides can be found in our user guide tutorial.
Pyreal's visualize module includes several functions that take in RealApp output directly to generate explanation plots.
All visualization functions take in many customization parameters. See the API reference for more information.
Feature Bar Plot
The feature bar plot can visualize general feature importance scores...
... or contribution scores for a single input
Strip Plot
Strip plots are an effective way to visualize feature contributions for multiple inputs at a time, to understand the general trends of how the ML model uses features.
To increase the amount of information displayed in these plots, you can generate feature contributions for the full training set.
Feature Scatter Plot
Scatter plots allow you to investigate how the model uses a specific feature, across the full range of that feature's values:
Example Table
To get a clean table comparing the feature values of a input data to those of similar examples, you can use the example_table
function:
This will give you a table like:
Last updated