USD/IDR Exchange Rate Prediction using Linear Regression

This repository contains a Jupyter Notebook that demonstrates how to predict the USD/IDR exchange rate using Linear Regression. The project leverages historical exchange rate data and applies machine learning techniques to forecast future exchange rates.

Overview

The project is divided into several key steps:

1. Data Collection: Historical USD/IDR exchange rate data from 2014 to 2024 is used. The data includes the date, open, high, low, and close exchange rates.

2. Data Preprocessing: The data is cleaned and preprocessed to make it suitable for machine learning. This includes converting the date format, aggregating monthly data, and preparing features for the model.

3. Model Training: A Linear Regression model is trained on the historical data. The model uses the year and month as features to predict the closing exchange rate.

4. Model Evaluation: The model's performance is evaluated using Root Mean Squared Error (RMSE) and R-squared metrics.

5. Future Predictions: The trained model is used to predict the USD/IDR exchange rate for the years 2025 to 2035.

6. Visualization: The results, including historical data and future predictions, are visualized using Matplotlib.

Requirements

To run the Jupyter Notebook, you need the following Python libraries:

• pyspark
• pandas
• numpy
• matplotlib

You can install the required libraries using pip:

pip install pyspark pandas numpy matplotlib

How to Use

1. Clone the Repository:

   git clone https://github.com/yourusername/LinearRegression_USDIDR.git
   cd LinearRegression_USDIDR

2. Run the Jupyter Notebook:

   jupyter notebook LinearRegression_USDIDR.ipynb

3. Follow the Steps in the Notebook:

   • The notebook is well-commented and guides you through each step of the process.
   • You can modify the code to experiment with different models or datasets.

Key Features

• Data Aggregation: The data is aggregated on a monthly basis to smooth out daily fluctuations and provide a clearer trend.
• Feature Engineering: The model uses the year and month as features, which are derived from the date.
• Model Evaluation: The model's performance is evaluated using RMSE and R-squared, providing insights into its accuracy.
• Future Predictions: The model predicts the exchange rate for the next decade, offering valuable insights for financial planning.

Results

The notebook includes visualizations that compare the historical exchange rates with the predicted values. These visualizations help in understanding the model's performance and the trends in the USD/IDR exchange rate.

Contributing

If you have any suggestions or improvements, feel free to open an issue or submit a pull request. Your contributions are welcome!

License

This project is licensed under the MIT License. See the LICENSE file for more details.

Acknowledgments

• The historical exchange rate data is sourced from a publicly available dataset.
• Special thanks to the developers of the libraries used in this project.

---

Feel free to explore the notebook and use the code to predict other financial metrics or adapt it for different datasets. Happy coding!