🟣Extracting Insights

Detecting Outliers and Edge Cases From Your Data

Get Insights From Your Data

Once you have completed your data setup, you can extract insights from your data by detecting outliers and edge cases.

This code snippet below generates Manot's insights on outliers that your model will perform poorly on. Once it has completed processing, Manot will return an ID value which we will use in the next step to visualize the insights Manot has identified.

insight = manot.insight(
    name="insight_example",
    setup_id=setup["id"],
    data_path="/path/to/data",
    data_provider="local",  # it can be "s3", "local" or "deeplake"
    percentage="percentage" # percentage of images to be considered insight should be larger than 0 and less or equal than 100
)

Visualize Your Data

Finally, we will use the ID generated in the previous step to visualize the outliers that Manot has identified.

insight_info = manot.get_insight(insight["id"])
manot.visualize_data_set(insight_info['data_set']['id'], deeplake_token)

Note: Providing a Deep Lake token in the line of code below is only required if you used Deep Lake as the data storage

Example of output:

To see a full end-to-end example, visit our Colab notebook.

Last updated