StockMarketPrediction

A comprehensive financial data analysis and machine learning optimization project. This repository implements and compares 9 different mathematical and machine learning models to analyze, forecast, and predict stock market pricing trends using historical data.

---

📖 Overview

Predicting stock market prices is a challenging time-series forecasting problem. This project processes historical data from the National Stock Exchange (NSE) to train and evaluate various predictive algorithms. By comparing advanced Deep Learning (LSTM), automated time-series forecasting (Auto-ARIMA, Prophet), and classical Machine Learning (Random Forest, SVM), this project attempts to find the most robust approach to estimate future price movements.

📊 Data Source & Characteristics

• Asset: Infosys (INFY)
• Exchange: NSE India (National Stock Exchange)
• Timeframe: July 1, 2015 – July 29, 2020
• Features Provided:
  • Date: Traded day.
  • Open / Close: Opening and closing price of the asset.
  • High / Low: Maximum and minimum price reached during the session.
  • Volume: Total number of shares traded.

🛠 Data Processing Pipeline

1. Cleaning: Missing (NaN) values detection and imputation/removal.
2. Date Transformation: Parsing temporal strings into standardized timestamp types (pd.to_datetime) for robust time-series indexing.
3. Feature Scaling: Normalization via Scikit-Learn's MinMaxScaler bound to (0,1) to stabilize gradient descent and proximity-based classifiers.
4. Data Splitting: Splitting timeline chronologically into Training Sets (historical patterns) to Validation Sets (future timeline).

🚀 Implemented Models

The repository evaluates 9 different algorithms, split by methodology:

Regression & Time-Series

• Linear Regression: A baseline statistical approach for mapping feature relations directly to prices.
• LSTM (Long Short-Term Memory): Advanced Deep Learning Recurrent Neural Network (RNN) tailored for time-series anomaly and pattern persistence.

Classification & Structuring Models

• Logistic Regression: Probabilistic trend classification.
• K-Nearest Neighbors (KNN): Regressive prediction mapping via the closest historical similarity states.
• Decision Tree: Non-linear, logical flowchart splits against data heuristics.
• Random Forest: Ensemble bagging framework using aggregated decision trees to drop variance and protect against overfitting.
• Support Vector Machines (SVM): High-dimensional mapping to separate complex financial coordinates.

📈 Evaluation Metrics

Model performance and ranking comparisons are formulated against:

• RMSE (Root Mean Squared Error): Standard deviation of prediction errors. Defines standard variance for regression fits.
• Accuracy: Base evaluation metric for trend prediction states (up/down).

⚙️ Setup & Installation

Follow these steps to initialize the environment:

1. Clone the repository:

   git clone https://github.com/arshsaxena/StockMarketPrediction.git
   cd StockMarketPrediction

2. Set up Virtual Environment (Recommended):

   python3 -m venv .venv
   source .venv/bin/activate  # On Windows use: .venv\Scripts\activate

3. Install Dependencies:

   pip install -r requirements.txt

4. Launch Jupyter Notebook Server:

   jupyter notebook

5. Begin Analysis: Open any specific model's .ipynb file to run cells interactively.

---

Note: All algorithms provided here are evaluated for academic and educational analytics purposes only and should not be deployed as independent financial trading advice.