tw.insights

The Insights class provides methods for extracting detailed insights from forecasting or anomaly detection models. It includes tools to analyze model properties and extract features, making it an essential tool for interpreting and understanding model behavior.

Class Initialization

To use the Insights API, access it through the TangentWorks class:

PY

from tangent_works import TangentWorks

# Initialize TangentWorks
tw = TangentWorks()

# Access the Insights class
tw.insights

Methods

properties

tw.insights.properties(model=Dict[str, Any])

Retrieves and processes the variable properties of the given model, sorted by importance.

Parameters

model (Dict[str, Any]): The model object containing variable properties.

Returns

pd.DataFrame: A DataFrame containing:
- Variable names.
- Importance scores, sorted in descending order.
- Relative importance scores.

Examples

PY

# Analyze model properties
properties_df = tw.insights.properties(model)

features

tw.insights.features(model=Dict[str, Any])

Extracts and processes the features used in the given model.

Parameters

model (Dict[str, Any]): The model object containing feature terms.

Returns

pd.DataFrame: A DataFrame containing:
- Model index.
- Term index.
- Feature name.
- Feature importance.
- Associated beta value.

Examples

PY

# Extract features from the model
features_df = tw.insights.features(model)

Key Features

Analyze model properties: Understand the importance of variables in your model using properties.
Feature extraction: Gain detailed insights into the features and terms used in your model with features.
Comprehensive results: Outputs are provided as easy-to-use Pandas DataFrames for further analysis and visualization.

Dependencies

pandas: Used for constructing and manipulating DataFrames.
Tangent Works core modules: Handle model processing and feature extraction.