Visualising predictions¶

Splink contains a variety of tools to help you visualise your predictions.

The idea is that, by developing an understanding of how your model works, you can gain confidence that the predictions it makes are sensible, or alternatively find examples of where your model isn't working, which may help you improve the model specification and fix these problems.

# Rerun our predictions to we're ready to view the charts
from splink.duckdb.linker import DuckDBLinker
from splink.datasets import splink_datasets
import altair as alt

df = splink_datasets.fake_1000
linker = DuckDBLinker(df)
linker.load_model("../demo_settings/saved_model_from_demo.json")
df_predictions = linker.predict(threshold_match_probability=0.2)

Waterfall chart¶

The waterfall chart provides a means of visualising individual predictions to understand how Splink computed the final matchweight for a particular pairwise record comparison.

To plot a waterfall chart, the user chooses one or more records from the results of linker.predict(), and provides these records to the linker.waterfall_chart() function.

For an introduction to waterfall charts and how to interpret them, please see this video.

records_to_view  = df_predictions.as_record_dict(limit=5)
linker.waterfall_chart(records_to_view, filter_nulls=False)

Comparison viewer dashboard¶

The comparison viewer dashboard takes this one step further by producing an interactive dashboard that contains example predictions from across the spectrum of match scores.

An in-depth video describing how to interpret the dashboard can be found here.

linker.comparison_viewer_dashboard(df_predictions, "scv.html", overwrite=True)

# You can view the scv.html file in your browser, or inline in a notbook as follows
from IPython.display import IFrame
IFrame(
    src="./scv.html", width="100%", height=1200
)

Cluster studio dashboard¶

Cluster studio is an interactive dashboards that visualises the results of clustering your predictions.

It provides examples of clusters of different sizes. The shape and size of clusters can be indicative of problems with record linkage, so it provides a tool to help you find potential false positive and negative links.

df_clusters = linker.cluster_pairwise_predictions_at_threshold(df_predictions, threshold_match_probability=0.5)

linker.cluster_studio_dashboard(df_predictions, df_clusters, "cluster_studio.html", sampling_method="by_cluster_size", overwrite=True)

# You can view the scv.html file in your browser, or inline in a notbook as follows
from IPython.display import IFrame
IFrame(
    src="./cluster_studio.html", width="100%", height=1200
)

Completed iteration 1, root rows count 11
Completed iteration 2, root rows count 1
Completed iteration 3, root rows count 0

Next steps¶

Now we have visualised the results of a model, we can move on to some more formal Quality Assurance procedures using labelled data.