FP&A Forecasting Model – Driver-Based + ML Approach

Author: Nicholas Hidalgo
Location: Boston, MA
Role: Business Intelligence & Analytics Leader
Tools: Python • Pandas • scikit-learn • Matplotlib • Tableau • Power BI • Snowflake • Databricks

---

Executive Summary

This project demonstrates an FP&A forecasting framework that integrates traditional driver-based forecasting with machine learning (Linear Regression) to enhance revenue accuracy and decision-making efficiency.

Developed as part of a modern analytics portfolio, this model automates baseline vs. ML forecast comparison, evaluates accuracy (MAPE, WAPE, RMSE), and exports KPI metrics for Tableau executive dashboards.

---

Project Structure

fpna_forecasting_model/ │ ├── data/ # Raw & processed datasets (CSV exports) ├── notebooks/ # Jupyter notebooks (modeling, accuracy, scenario analysis) ├── docs/ # Visuals & documentation (for README & Tableau) ├── dashboard/ # Tableau dashboard assets ├── README.md # Project documentation └── requirements.txt # Environment dependencies

---

Objectives

• Develop a baseline & ML forecasting model for revenue prediction
• Quantify accuracy using MAPE, WAPE, and RMSE
• Export standardized KPI metrics for Tableau
• Visualize results for executive reporting
• Build future scenario & sensitivity analysis (next phase)

---

Key Model Metrics (Sample Output)

Model	MAPE	WAPE	RMSE
Baseline (Last Value)	6.9%	7.3%	3,210
ML Linear Regression	4.1%	4.5%	2,540

Metrics calculated using test set from baseline vs. ML forecast comparison.

---

Forecast Visualization

A visual comparison of actual vs. forecasted revenue using a driver-based and ML model.

Interpretation

• Baseline: Holds the last observed revenue constant.
• ML model: Predicts revenue using time, volume, and price drivers.
• ML model tracks actuals more closely → higher predictive accuracy.
• Metrics exported to data/model_metrics.csv for Tableau dashboards.

---

How to Run

1. Clone the repository:

   git clone https://github.com/nicholasjh-work/fpna-forecasting-model.git
   cd fpna-forecasting-model
   

---

Objectives

• Develop a baseline & ML forecasting model for revenue prediction
• Quantify accuracy using MAPE, WAPE, and RMSE
• Export standardized KPI metrics for Tableau
• Visualize results for executive reporting
• Build future scenario & sensitivity analysis (next phase)

---

Key Model Metrics (Sample Output)

Model	MAPE	WAPE	RMSE
Baseline (Last Value)	6.9%	7.3%	3,210
ML Linear Regression	4.1%	4.5%	2,540

Metrics calculated using test set from baseline vs. ML forecast comparison.

---

Forecast Visualization

A visual comparison of actual vs. forecasted revenue using a driver-based and ML model.

Interpretation

• Baseline model: Holds the last observed revenue constant.
• ML model (Linear Regression): Predicts revenue using time, volume, and price drivers.
• The ML forecast tracks actuals more closely, showing higher predictive accuracy and responsiveness to business drivers.
• Exported metrics (MAPE, WAPE, RMSE) are stored in data/model_metrics.csv for Tableau or Power BI dashboards.

---

Portfolio Links

Platform	Link
GitHub	nicholasjh-work
Tableau Public	Nicholas Hidalgo
Kaggle	nicholasjhidalgo
LinkedIn	nicholasjhidalgo

---

⚙️ How to Run Locally

Follow these steps to reproduce the forecasting results and generate outputs for Tableau or Power BI.

Clone the repository

git clone https://github.com/nicholasjh-work/fpna-forecasting-model.git
cd fpna-forecasting-model

python -m venv venv
source venv/bin/activate   # On macOS / Linux
venv\Scripts\activate      # On Windows

pip install -r requirements.txt

jupyter lab

notebooks/02_accuracy_reporting.ipynb

---

This notebook will:

Generate baseline and ML forecasts Compute accuracy metrics (MAPE, WAPE, RMSE) Export outputs to data/model_metrics.csv Save forecast visualization to docs/actuals_vs_forecasts.png

Optional: Explore in Tableau

You can connect Tableau to: data/model_metrics.csv for KPI metrics data/synthetic_financials.csv for scenario visualization

These datasets power executive dashboards for FP&A insights.

---