Supply Chain Emissions Modeling Using Industry and Commodity Data (2010–2016) Problem Statement:

You have annual supply chain emission data from 2010–2016 categorized into industries and commodities. The goal is to develop a regression model that can predict the Supply Chain Emission Factors with Margins based on descriptive and quality metrics (substance, unit, reliability, temporal/geographical/technological/data collection correlations, etc.).

📊 Supply Chain Greenhouse Gas (GHG) Emissions Prediction Project

---

🎯 Project Goal

To analyze and predict greenhouse gas (GHG) emissions from various U.S. industries and commodities using official supply chain data sourced from data.gov.

The aim is to build robust regression models that can predict GHG Emissions (kg CO₂ equivalent) based on industry, commodity, and quality-related descriptive features.

---

🌐 Data Source

• Dataset Name: Supply Chain Greenhouse Gas Emission Factors
• Source: data.gov
• Years Covered: 2010 – 2016

---

📂 Dataset Overview

The dataset contains annual supply chain GHG emission factors across various U.S. industries and commodities.

---

🧹 Data Preprocessing Workflow

1️⃣ Data Loading

• Multiple Excel sheets loaded from 2010–2016.
• Combined Industry and Commodity data into one dataset.

2️⃣ Data Cleaning

• Handle missing values (drop or impute).
• Convert data types to appropriate formats.
• Standardize column names.
• Remove duplicate rows.

3️⃣ Feature Engineering

• Encode categorical features (e.g., Industry_Name, Commodity, Substance, Source).
• Normalize/scale numeric columns for model compatibility.

---

🔎 Exploratory Data Analysis (EDA)

• Analyzed distributions of GHG emissions.
• Investigated relationships between descriptive metrics and emission factors.
• Detected outliers and patterns across industries and commodities.

---

🤖 Model Building & Evaluation

✅ Target Variable

• GHG_Emissions_kgCO2e

✅ Features Used

• Industry & Commodity Codes
• Reliability and Correlation scores
• Year, Source, Substance, Units (encoded)

✅ Models Applied

• Linear Regression (Baseline)
• Random Forest Regression
• (Optional: Other regressors like XGBoost, SVR for comparison)

✅ Evaluation Metrics

Metric	Description
RMSE	Root Mean Squared Error
MAE	Mean Absolute Error
R² Score	Coefficient of Determination

---

🔧 Tools & Technologies Used

• Language: Python 3.x
• Libraries:
  • Pandas
  • NumPy
  • Scikit-learn
  • Matplotlib
  • Seaborn

---

🔬 Modeling Workflow

1️⃣ Step 1: Import Required Libraries
2️⃣ Step 2: Load Dataset
3️⃣ Step 3: Data Preprocessing (Cleaning + EDA + Encoding)
4️⃣ Step 4: Split Dataset (Train-Test Split)
5️⃣ Step 5: Model Training
6️⃣ Step 6: Prediction and Evaluation
7️⃣ Step 7: Hyperparameter Tuning
8️⃣ Step 8: Comparative Study and Best Model Selection

---

📌 Author

Satyam Mishra

---