https://github.com/raghavendranhp/industrial_copper_modelling
Industrial Copper Modeling optimizes pricing decisions using advanced ML. Predict sales with accuracy, classify leads, and streamline decision-making.
https://github.com/raghavendranhp/industrial_copper_modelling
classification-models copper decision-tree-classifier decision-tree-regression pickle-file predictive-modeling regression-models scikit-learn
Last synced: 21 days ago
JSON representation
Industrial Copper Modeling optimizes pricing decisions using advanced ML. Predict sales with accuracy, classify leads, and streamline decision-making.
- Host: GitHub
- URL: https://github.com/raghavendranhp/industrial_copper_modelling
- Owner: raghavendranhp
- Created: 2023-11-30T03:55:16.000Z (over 2 years ago)
- Default Branch: main
- Last Pushed: 2024-01-12T05:50:14.000Z (over 2 years ago)
- Last Synced: 2025-01-07T06:33:36.274Z (over 1 year ago)
- Topics: classification-models, copper, decision-tree-classifier, decision-tree-regression, pickle-file, predictive-modeling, regression-models, scikit-learn
- Language: Jupyter Notebook
- Homepage:
- Size: 17 MB
- Stars: 0
- Watchers: 1
- Forks: 0
- Open Issues: 0
-
Metadata Files:
- Readme: README.md
Awesome Lists containing this project
README
# Industrial Copper Modeling
### Domain: Manufacturing
🔗 **Live App**: [Streamlit App](https://industrialcoppermodelling-ragha22.streamlit.app/)
---
## 📌 Problem Statement
The copper industry often grapples with noisy, skewed sales and pricing data, making manual prediction and analysis inefficient and inaccurate. Additionally, the challenge of identifying quality leads based on sales status adds complexity to decision-making. This project aims to:
* Build a **Regression Model** to predict the continuous variable `Selling_Price`.
* Develop a **Classification Model** to predict lead conversion status (`WON` or `LOST`).
* Create an **interactive Streamlit web application** to allow users to input features and get real-time predictions for either task.
---
## 🗂️ Dataset Description
The dataset contains information on copper product sales including transactional, customer, and product specifications.
| Feature | Description |
| --------------- | ---------------------------------------------------- |
| `id` | Unique identifier for each record |
| `item_date` | Transaction date |
| `quantity tons` | Quantity sold (in tons) |
| `customer` | Customer identifier |
| `country` | Customer country |
| `status` | Status of deal (e.g., WON, LOST) |
| `item type` | Type of item sold |
| `application` | Application code for item |
| `thickness` | Thickness of the item |
| `width` | Width of the item |
| `material_ref` | Material reference code |
| `product_ref` | Product reference code |
| `delivery date` | Delivery date |
| `selling_price` | Selling price of the product (Target for Regression) |
---
## 🔍 Approach
### 1. Data Understanding
* Identified variable types (categorical/continuous)
* Treated anomalies in `material_ref` (e.g., values starting with `00000`)
* Handled reference columns as categorical features
### 2. Data Preprocessing
* Imputed missing values using statistical techniques (mean/median)
* Handled outliers via **IQR method** and **Isolation Forest**
* Treated skewness with **log transformation** and **Box-Cox**
* Encoded categorical variables using **One-Hot** and **Label Encoding**
### 3. Exploratory Data Analysis (EDA)
* Visualized skewness and outliers using `Seaborn` (boxplot, violinplot)
* Correlation heatmaps to drop highly correlated features
### 4. Feature Engineering
* Created new informative features when needed
* Dropped redundant or highly correlated features
### 5. Model Building
#### Regression:
* Tree-based models (Random Forest, XGBoost) used due to data non-linearity and noise
* Target: `selling_price`
* Evaluation Metrics: MAE, RMSE, R² Score
#### Classification:
* Models: Logistic Regression, ExtraTreesClassifier, XGBClassifier
* Target: `status` (filtered for `WON`, `LOST`)
* Evaluation Metrics: Accuracy, Precision, Recall, F1-score, AUC-ROC
### 6. Streamlit Application
* Choose task: Regression or Classification
* Input fields for all feature columns (excluding target)
* Model predictions displayed instantly
* Behind the scenes:
* Applied same preprocessing (scaling, encoding, transformation)
* Loaded trained models using `pickle`
---
## ✅ Conclusion
This end-to-end machine learning solution offers real-time insights into copper product pricing and lead conversion. By automating predictions and deploying them via a simple web app, the project supports better decision-making in the manufacturing domain.
## Getting Started
To run the project locally, follow these steps:
1. Clone the repository:
```bash
git clone https://github.com/raghavendranhp/Industrial_copper_modelling.git
cd industrial-copper-modeling
```
2. Install the required dependencies:
```bash
pip install -r requirements.txt
```
3. Execute the main script:
```bash
python main_script.py
```
4. Access the Streamlit web application by opening the provided URL in your browser.
## Project Structure
```
industrial-copper-modeling/
│
├── data/ # Directory for storing datasets
├── models/ # Directory for saving trained models
├── Copper_Modelling.ipynb # Jupyter notebooks for exploration and analysis
├── .gitignore
├── main_script.py # Main script for running the project
├── requirements.txt # List of project dependencies
└── README.md # Project documentation
```
## Contributions
Contributions to the project are welcome. If you find any issues or have suggestions for improvement, feel free to open an issue or submit a pull request.
## License
This project is licensed under the [MIT License](LICENSE).