📈 Financial Time Series Modeling

This project applies a range of statistical and machine learning models to analyze and forecast financial time series data such as log returns and realized volatility. Both R and Python are used in the workflow.

🗂 Project Structure

File	Description
Preprocessing.R	Cleans and visualizes log return data for selected stocks.
ARIMA.R	Fits ARIMA models on log return series and generates out-of-sample forecasts.
Stationarity Check.R	Performs ADF test and ACF/PACF plotting to validate model assumptions.
Volatility Forecasting.R	Models realized volatility using ARIMA and compares predictions with test data.
LASSO.R	Implements LASSO regression to predict volatility with regularization.
time_series_lstm.py	Builds and trains an LSTM neural network to model volatility trends.
README.md	Project documentation.

📊 Methods Used

• Linear regression
• ARIMA and time series diagnostics (ACF, PACF, ADF)
• LASSO regression (with glmnet)
• LSTM (Long Short-Term Memory) neural network (with TensorFlow)

🧪 How to Run

R Scripts

Make sure the following R packages are installed:

install.packages(c("forecast", "tseries", "zoo", "xts", "glmnet", "caret", "ggplot2", "readr", "dplyr"))

You can knit each .R script in RStudio or run line-by-line interactively.

Python Script

Ensure you have the required packages:

pip install numpy pandas matplotlib scikit-learn tensorflow

Then run:

python time_series_lstm.py

📁 Data

All scripts assume cleaned datasets are stored under a data/ folder. Example filenames:

• updated_data.csv
• aal_forecast_data.csv
• updated_volatility_data.csv
• lasso_data.csv
• 222_scaled.csv

(Replace or modify paths as needed in your local setup.)

This repository combines interpretable models with modern deep learning for a comprehensive analysis of financial time series.