📁 mmm¶

The mmm (Marketing Mix Modeling) module provides a comprehensive suite of tools for building, analyzing, and visualizing marketing mix models. It is organized into several submodules, each responsible for a specific aspect of the modeling workflow.

Submodules¶

analysis¶

The analysis submodule contains tools for evaluating and interpreting model results.

classifier.py: Implements classification utilities for model outputs.
contribution.py: Calculates the contribution of each feature or channel.
metrics.py: Provides metrics for model evaluation.
response_curve.py: Generates and analyzes response curves.
s_curve_filler.py: Fills and smooths S-curves for response analysis.
summary.py: Summarizes model results and key findings.

config¶

The config submodule manages configuration settings for MMM pipelines.

transform_config.py: Handles transformation configurations for data preprocessing.

models¶

The models submodule defines the core modeling classes and interfaces.

base.py: Abstract base classes for all models.
liner.py: Linear model implementations.
simple_model.py: Simple baseline models for quick prototyping.
sklearn.py: Wrappers and utilities for scikit-learn models.

pipeline¶

The pipeline submodule orchestrates the end-to-end modeling workflow.

model_pipeline.py: Defines the main pipeline for model training and evaluation.
pipeline.py: General pipeline utilities and helpers.

transformers¶

The transformers submodule provides data transformation utilities.

adstock.py: Implements adstock transformation for media variables.
base.py: Base classes for all transformers.
hill.py: Hill function transformation for diminishing returns.
log.py: Log transformation utilities.
saturation.py: Saturation transformation for media response.
scaler.py: Scaling utilities for feature normalization.

utils¶

The utils submodule contains helper functions and utilities.

helpers.py: General-purpose helper functions used throughout the module.

visualization¶

The visualization submodule provides tools for visualizing model results.

marginal_roi.py: Plots marginal ROI curves.
plotter.py: General plotting utilities for model outputs.

Usage Example¶

from owlmix.mmm.models.simple_model import SimpleModel
from owlmix.mmm.pipeline.model_pipeline import ModelPipeline
from owlmix.mmm.transformers.adstock import AdstockTransformer

# Example: Build and run a simple MMM pipeline
model = SimpleModel()
transformer = AdstockTransformer()
pipeline = ModelPipeline(model=model, transformer=transformer)
results = pipeline.run(data)

# Visualize results
from owlmix.mmm.visualization.plotter import plot_results
plot_results(results)

Module Structure¶

mmm/
├── analysis/
├── config/
├── models/
├── pipeline/
├── transformers/
├── utils/
└── visualization/

For detailed API documentation, refer to the respective submodule documentation.