Data

Tools

Indexes

Applications

Experiments

Awesome

An open API service indexing awesome lists of open source software.

https://github.com/karan-sharma-aiml/future_ml_01

Hands-on Machine Learning project built with Python, Streamlit, Prophet & TensorFlow | Showcasing AIML skills through time-series forecasting & real-world data insights.
https://github.com/karan-sharma-aiml/future_ml_01

aiml data-visualisation machine-learning project prophet python streamlit tensorflow timreactnative

Last synced: about 2 months ago
JSON representation

Hands-on Machine Learning project built with Python, Streamlit, Prophet & TensorFlow | Showcasing AIML skills through time-series forecasting & real-world data insights.

Awesome Lists containing this project

README 

          
# 🚀 AI-Powered Sales Forecasting Dashboard — Advanced



[![Python](https://img.shields.io/badge/Python-3.8+-blue.svg)](https://python.org)

[![Streamlit](https://img.shields.io/badge/Streamlit-1.28+-red.svg)](https://streamlit.io)

[![Prophet](https://img.shields.io/badge/Prophet-1.1+-green.svg)](https://facebook.github.io/prophet/)

[![TensorFlow](https://img.shields.io/badge/TensorFlow-2.12+-orange.svg)](https://tensorflow.org)



> **Advanced Time Series Forecasting Dashboard with Multi-Model Ensemble, Uncertainty Quantification, and Explainable AI**



A production-ready forecasting system that combines **Prophet**, **LSTM Neural Networks**, and **XGBoost** with comprehensive model explainability using **SHAP**.



---



## 📊 Dashboard Preview



### 🏠 Main Dashboard  



![Dashboard Overview](./screenshots/1.Dashbard%20section/01_dashbard_overview.png)



Uploaded sales data preview in real time



---



### 🔮 Forecasting Models  

![Prophet Forecast](screenshots/2.Forecasting%20Section/04_prophet_forecast.png)  

*Prophet model prediction with seasonality components*  



![LSTM Forecast](screenshots/2.Forecasting%20Section/06_lstm_forecast.png)  

*LSTM Neural Network with MC Dropout based uncertainty*  



![Combined Models](screenshots/2.Forecasting%20Section/07_combined_prophet_lstm.png)  

*Comparison of Prophet & LSTM predictions*



---



### 📈 Model Evaluation & Explainability  

![Rolling Backtest](screenshots/3.Evaluation%20&%20Explainability/08_rolling_bracket_table.png)  

*Rolling-origin cross-validation metrics (MAE, RMSE, MAPE)*  



###  SHAP Bar Chart  

![SHAP Bar Chart](./screenshots/3.Evaluation%20%26%20Explainability/09_Sharp_bar_chart.png)



### Feature Importance

![Feature Importance](screenshots/3.Evaluation%20&%20Explainability/10_feature_importance.png)  

*XGBoost feature contributions visualized*



```markdown

## 📂 Project Structure - FUTURE_ML_01



📂 FUTURE_ML_01/ - Root project folder

├─ 📂 data/ - Contains dataset files

│  └─ 📄 sample_sales.csv - Sample sales dataset

├─ 📂 notebook/ - Jupyter notebooks for analysis

│  ├─ 📓 eda.ipynb - Exploratory Data Analysis

│  └─ 📓 model_experiments.ipynb - Model experimentation

├─ 📂 screenshots/ - Screenshots of dashboard and results

│  ├─ 📂 1.Dashboard Section/ - Dashboard UI images

│  │  ├─ 🖼️ 01_dashboard_overview.png - Overview of dashboard

│  │  ├─ 🖼️ 02_sidebar_controls.png - Sidebar controls screenshot

│  │  └─ 🖼️ 03_data_preview.png - Sample data preview

│  ├─ 📂 2.Forecasting Section/ - Forecasting model outputs

│  │  ├─ 🔮 04_prophet_forecast.png - Prophet forecast chart

│  │  ├─ 📅 05_prophet_components_main.png - Prophet main components

│  │  ├─ 📆 05_prophet_yearly.png - Yearly trend components

│  │  ├─ 🤖 06_lstm_forecast.png - LSTM forecast chart

│  │  └─ ⚡ 07_combined_prophet_lstm.png - Combined forecast

│  └─ 📂 3.Evaluation & Explainability/ - Evaluation visuals

│     ├─ 📋 08_rolling_bracket_table.png - Rolling bracket table

│     ├─ 📊 09_Sharp_bar_chart.png - Sharpe ratio chart

│     └─ 🌟 10_feature_importance.png - Feature importance chart

├─ 📂 src/ - Source code for models and utilities

│  ├─ ⚙️ backtesting.py - Backtesting logic

│  ├─ ⚙️ data_processing.py - Data cleaning and preprocessing

│  ├─ ⚙️ lstm_model.py - LSTM model implementation

│  ├─ ⚙️ prophet_model.py - Prophet model implementation

│  ├─ ⚙️ utils.py - Utility functions

│  └─ ⚙️ xgb_baseline.py - XGBoost baseline model

├─ ⚙️ .gitignore - Git ignore file

├─ 📜 LICENSE - License file

├─ 📘 README.md - Project documentation

├─ 📦 requirements.txt - Python dependencies

└─ 🌐 streamlit_sales_forecast.py - Main Streamlit app



## 🌟 Key Features

- Upload custom sales CSV files (time series with date and sales columns)  

- Automatic feature engineering: lag features, rolling averages, and seasonality indicators  

- Multi-model sales forecasting using Prophet, LSTM (with MC dropout uncertainty), and XGBoost  

- Visualize model forecasts with confidence intervals and trend/seasonality decomposition  

- Rolling-origin cross-validation with performance metrics such as MAE, RMSE, MAPE, and coverage  

- Explainability using SHAP for model interpretability  

- Interactive Streamlit dashboard with real-time parameter tuning



### 🔬 **Advanced ML Models**

- **Facebook Prophet** with automatic seasonality detection and uncertainty intervals

- **LSTM Neural Networks** with Monte Carlo Dropout for uncertainty quantification  

- **XGBoost Baseline** with feature importance and SHAP explainability

- **Multi-model ensemble** with rolling-origin cross-validation



### 📊 **Professional Dashboard**

- **Interactive Streamlit interface** with real-time parameter tuning

- **Dynamic visualization** with Plotly charts

- **Comprehensive data preview** with automated feature engineering

- **Export functionality** for predictions and plots



### 🎯 **Production-Ready Architecture**

- **Modular codebase** with separation of concerns

- **Testing suite** with pytest

- **Config management** with YAML

- **Professional docs** & clean organization



---



## 🛠️ Technical Architecture



### **Pipeline**

```



Data Ingestion → Feature Engineering → Model Training →

Prediction Generation → Model Validation → Explainability Analysis



````



### **Tech Stack**

- **Backend**: Python 3.8+ (Pandas, NumPy, scikit-learn ecosystem)

- **ML Frameworks**: TensorFlow/Keras, Prophet, XGBoost

- **Explainability**: SHAP

- **Visualization**: Plotly

- **UI**: Streamlit



---



## 🚀 Quick Start



```bash

# Clone repo

git clone https://github.com/karan-sharma-aiml/FUTURE_ML_01.git

cd FUTURE_ML_01



# Setup env

python -m venv venv

source venv/bin/activate   # Windows: venv\Scripts\activate



# Install dependencies

pip install -r requirements.txt



# Run dashboard

streamlit run streamlit_sales_forecast.py

````



---



## 📊 Model Performance (Example)



| Model   | MAE   | RMSE  | MAPE  | Coverage% | Training Time |

| ------- | ----- | ----- | ----- | --------- | ------------- |

| Prophet | 0.00  | 0.00  | 0.00% | 95.2%     | 2.3s          |

| LSTM    | 0.00  | 0.00  | 0.00% | 92.8%     | 45.7s         |

| XGBoost | 70.03 | 85.21 | 12.4% | -         | 1.8s          |



---



## 🎯 Business Applications



* **Retail Forecasting**: Inventory & demand planning

* **Revenue Forecasting**: Budget allocation

* **Resource Planning**: Workforce scheduling

* **Supply Chain Optimization**: Procurement & logistics



---



## 🤝 Contributing



Contributions are welcome!



1. Fork the repo

2. Create a feature branch

3. Commit changes + add docs/tests

4. Submit PR 🎉



---



## 👨‍💻 Author



**Karan Sharma**

🎓 B.Tech CSE (AI/ML) Student @ CGC University, Mohali



💡 Focus: Time Series | Deep Learning | Explainable AI

📧 Email: [karan.sharma@email.com](mailto:karan.sharma@email.com)

🔗 [LinkedIn](https://www.linkedin.com/in/karan-sharma-167957271)

🐙 [GitHub](https://github.com/karan-sharma-aiml)



---






⭐ **Star this repo if you find it useful!** ⭐



*Built with ❤️ for the AI/ML community*





```



---