https://github.com/jra333/marginal-cpc-calculator

https://github.com/jra333/marginal-cpc-calculator

Last synced: 4 days ago
JSON representation

• Host: GitHub
• URL: https://github.com/jra333/marginal-cpc-calculator
• Owner: jra333
• Created: 2025-02-06T08:13:28.000Z (8 months ago)
• Default Branch: main
• Last Pushed: 2025-02-07T02:59:29.000Z (8 months ago)
• Last Synced: 2025-05-19T09:11:51.533Z (5 months ago)
• Language: Jupyter Notebook
• Size: 32.7 MB
• Stars: 0
• Watchers: 1
• Forks: 0
• Open Issues: 0
• Metadata Files:
  • Readme: README.md

Awesome Lists containing this project

README

          
# Marginal CPC Analysis

This repository contains tools to analyze and forecast Cost Per Click (CPC) metrics for digital advertising campaigns. It includes two main components: a robust data processing and modeling notebook and an interactive Streamlit app for real-time CPC predictions based on future spend.

---

## Overview

### 1. [CPC Modeling Notebook](https://github.com/jra333/marginal_dimreturns_cpc/tree/main/marginal_cpc_notebooks)

- **Data Extraction & Cleaning**

  - Connects to Snowflake to retrieve raw cost, clicks, impressions, and search impression share data.

  - Cleans and preprocesses the data (e.g., date conversion, type corrections, sorting, and re-indexing).

  

- **Feature Engineering & Aggregation**

  - Computes additional variables like CPC and Eligible Impressions.

  - Applies resampling (daily) and aggregation.

  - Creates lag features and rolling averages for deeper time-series insights.

  

- **Exploratory Analysis**

  - Splits the dataset by year (2022, 2023, 2024) and provides descriptive statistics.

  - Exports the cleaned dataset for further analysis.

- **Modeling**

  - Leverages the cleaned and engineered dataset to train regression-based models aimed at forecasting CPC.

  - Performs model fitting, validation, and selection using various techniques (e.g., Linear Regression, Ridge, Lasso, etc.).

  - Provides diagnostic metrics (e.g., R², MSE, MAE) to assess model performance.

  - Saves the best performing model (with scalers and feature names) that is later used within the interactive app for real-time predictions.

### 2. [Interactive CPC Prediction App](https://github.com/jra333/marginal_dimreturns_cpc/tree/main/marginal_cpc_app_testing)

- **Model Loading & Prediction**

  - Loads a pre-trained model along with relevant scalers and feature names from a pickle file.

  - Provides a prediction function that uses scaled inputs to forecast the CPC.

  

- **Data Upload & Model Update**

  - Users can upload a CSV file of historical data that is parsed and cleaned inside the app.

  - If new data is present, lag features are generated and the model is updated (re-trained) accordingly.

  

- **User Interface**

  - Offers sliders for entering daily spend values for a week.

  - Computes and displays:

    - Predicted CPC per day.

    - Estimated number of clicks (Spend/CPC).

    - Estimated impressions (using user-defined CTR).

  - Features interactive visualizations (via Plotly) showing spend versus predicted CPC along with a regression trendline.