📁 analysis.classifier

The ResponseCurveClassifier module within the analysis submodule of the MMM package provides tools for classifying model results based on predefined criteria. This can be useful for categorizing model performance, identifying key drivers, or segmenting results for further analysis.

Overview

The module exposes a classifier class that:

• Accepts model results and classification parameters

• Applies classification rules to categorize results

• Returns structured output for downstream analysis or reporting

Class Reference

class mmm.analysis.classifier.ResponseCurveClassifier(curve, low_ratio=0.3, high_ratio=0.7)

Classifies model results based on specified criteria.

Parameters:

• curve (Dict) – Configuration parameters for classification.

• low_ratio (float) – Threshold for classifying results as “underspend”. Default is 0.3.

• high_ratio (float) – Threshold for classifying results as “saturated”. Default is 0.7.

classify()

Applies classification rules to the provided model results.

Returns:

A dictionary containing classified results.

Return type:

Dict

Result Example

{
  "zones": [
    "underspend", "underspend", ...,
    "optimal", "optimal", ...,
    "saturated", "saturated", ...
  ],
  "marginal": [0.2, 0.4, ..., 0.5, 0.6, ..., 0.3, 0.1],
  "thresholds": {
    "low": 0.3,
    "high": 0.7
  }
}

Notes

• The classification logic is based on predefined thresholds for marginal returns, which can be customized as needed.

• This module is designed to work with the output of MMM models, providing insights into the effectiveness of different marketing channels or tactics.

• The results can be used to inform budget allocation decisions, optimize marketing strategies, or identify areas for improvement in the model.

References

• MMM Overview

Back to Home