An end-to-end financial analytics and machine learning project that analyzes historical performance of NIFTY stocks, predicts next-day stock price movement using technical indicators, and presents the results through an interactive Streamlit dashboard.
🌐 Live Demo:
https://sector-wise-performance-and-risk-analysis-ml.streamlit.app/
Disclaimer:
This project is for educational purposes only and should not be considered financial or investment advice.
- 📥 Automated historical stock data collection using Yahoo Finance
- 📊 Statistical analysis of daily stock returns
- 📈 Annual return and volatility calculations
- ⚖️ Sharpe Ratio computation for risk-adjusted performance
- 🏦 Sector-wise performance comparison
- 🔗 Correlation analysis between stocks
- 📉 Professional financial visualizations
- ⚙️ Technical indicator generation
- 🤖 Machine Learning pipeline for stock movement prediction
- 🧠 Automatic model comparison and best model selection
- 💾 Model persistence using Joblib
- 🔮 Next-day stock movement prediction with confidence scores
- 📄 Automatic CSV report generation
- Analysed 24 NIFTY stocks across 5 sectors
- Built an end-to-end ML pipeline
- Engineered 13 technical indicators
- Trained and compared multiple classification models
- Developed an interactive multi-page Streamlit dashboard
- Automated prediction generation for all supported stocks
Historical Stock Data
│
▼
Data Collection (Yahoo Finance)
│
▼
Data Cleaning & Processing
│
▼
Financial Metrics Calculation
│
▼
Visualization Generation
│
▼
Feature Engineering
│
▼
Machine Learning Models
│
▼
Model Evaluation
│
▼
Best Model Selection
│
▼
Prediction Generation
│
▼
Interactive Streamlit Dashboard
Sector-Performance-Analysis-ML/
│
├── assets/ # Images used in the README
│
├── dashboard/ # Reusable Streamlit components
│ ├── chart_style.py
│ ├── charts.py
│ ├── components.py
│ ├── config.py
│ ├── indicators.py
│ ├── load_data.py
│ ├── sidebar.py
│ ├── styles.py
│ ├── technical_indicators.py
│ └── theme.py
│
├── data/
│ ├── raw/ # Historical stock data
│ └── processed/ # Processed datasets
│
├── figures/ # Generated visualizations
│
├── models/ # Trained ML models
│
├── outputs/
│ ├── model_summary.csv
│ ├── stock_summary.csv
│ ├── descriptive_statistics.csv
│ └── predictions/ # Model prediction outputs
│
├── pages/ # Streamlit application pages
│ ├── 1_Dashboard.py
│ ├── 2_Sector_Analysis.py
│ ├── 3_Stock_Analysis.py
│ ├── 4_Model_Analysis.py
│ ├── 5_Prediction.py
│ └── 6_About.py
│
├── src/
│ ├── config.py
│ ├── data_ingestion.py
│ ├── feature_engineering.py
│ ├── metrics.py
│ ├── visualize.py
│ └── ml/
│ ├── evaluation.py
│ ├── ml_pipeline.py
│ ├── model_training.py
│ ├── prediction.py
│ ├── prediction_pipeline.py
│ ├── preprocessing.py
│ └── utils.py
│
├── app.py # Streamlit entry point
├── main.py # Data processing pipeline
├── requirements.txt
├── LICENSE
└── README.md
The project analyzes 24 large-cap NIFTY stocks across five major sectors:
| Sector | Companies |
|---|---|
| Banking | HDFC Bank, ICICI Bank, SBI, Axis Bank, Kotak Bank |
| Information Technology | TCS, Infosys, Wipro, HCLTech, Tech Mahindra |
| Pharmaceuticals | Sun Pharma, Dr. Reddy's, Cipla, Divi's Labs, Lupin |
| Automobile | Maruti Suzuki, M&M, Bajaj Auto, Hero MotoCorp |
| FMCG | Hindustan Unilever, ITC, Nestlé India, Britannia, Dabur |
Historical data is downloaded automatically using the yfinance API.
The project computes:
- Daily Returns
- Descriptive Statistics
- Annual Returns
- Annual Volatility
- Sharpe Ratio
- Correlation Matrix
- Sector Performance
- Investment Growth
The machine learning pipeline creates the following technical indicators:
| Feature |
|---|
| Daily Return |
| 5-Day Moving Average |
| 10-Day Moving Average |
| 20-Day Moving Average |
| Volume Change |
| Rolling Volatility |
| High-Low Percentage |
| Open-Close Percentage |
| RSI (Relative Strength Index) |
| MACD |
| MACD Signal |
| Bollinger Band Upper |
| Bollinger Band Lower |
These features are used to predict whether the stock price will move UP or DOWN on the following trading day.
Three supervised learning algorithms are trained and evaluated.
| Model | Purpose |
|---|---|
| Logistic Regression | Baseline linear classifier |
| Decision Tree | Non-linear decision model |
| Random Forest | Ensemble learning model |
Each model is evaluated using:
- Accuracy
- Precision
- Recall
- F1 Score
The model with the highest F1 Score is automatically selected and saved.
The project automatically generates:
- Risk vs Return Scatter Plot
- Correlation Heatmap
- Sector-wise Returns
- Sector-wise Volatility
- Investment Growth Comparison
- Top 5 Performing Stocks
- Bottom 5 Performing Stocks
- Sector Growth Comparison
- Sharpe Ratio Comparison
All plots are saved inside the figures/ directory.
After execution, the project automatically creates:
outputs/
│
├── descriptive_statistics.csv
├── stock_summary.csv
├── model_summary.csv
It also generates:
- trained machine learning models
- prediction CSV files
- processed datasets
- financial visualizations
git clone https://github.com/abhiminav/Sector-Performance-Analysis-ML.git
cd Sector-Performance-Analysis-ML
pip install -r requirements.txtpython main.pystreamlit run app.py| Category | Technology |
|---|---|
| Programming | Python |
| Data Analysis | Pandas, NumPy |
| Machine Learning | Scikit-learn |
| Dashboard | Streamlit |
| Interactive Charts | Plotly |
| Static Visualisation | Matplotlib |
| Data Source | yfinance |
| Model Persistence | Joblib |
Potential enhancements include:
- Cross-validation
- Hyperparameter tuning using GridSearchCV
- XGBoost and LightGBM models
- LSTM-based time-series forecasting
- Portfolio optimization
- Real-time stock prediction
- Model explainability using SHAP values
Abhinav Mishra
B.Tech Computer Science Engineering
Built as a portfolio project showcasing skills in:
- Data Analysis
- Financial Analytics
- Machine Learning
- Streamlit Dashboard Development
This project is licensed under the MIT License.