This project involves implementing a sequential decision-making process for investment in Gold, Stock, and Bond. The decision-making process is conducted weekly, utilizing historical data and a linear regression model for future predictions. The integration of linear programming using MiniZinc ensures interpretable objectives and constraints.
- Ali Maher
- Mehdi Vakili
In this part, a linear regression model is implemented for fitting a line to the data points representing object prices and dates. Due to solver limitations in MiniZinc, challenges were faced in handling float variables and implementing absolute functions. A workaround involved using constraints to mimic the behavior of an absolute function.
The second part focuses on implementing a MiniZinc file for decision-making, optimizing the profit from investments in gold, stock, and bonds based on predicted data. Constraints are applied to ensure the accuracy of the calculated budget.
The final part involves data handling using NumPy and Pandas, time handling with the DateTime library, connecting the prediction and decision-making parts, and showcasing results with visualizations. A Streamlit app is developed for an interactive and user-friendly display of weekly plots, decisions, and budget.
- Install required dependencies:
pip install -r requirements.txt - Run the Streamlit app:
streamlit run main.py
The project provides insights into short-term investment strategies, suggesting the optimal time window for decision-making. The GitHub repository contains code, documentation, and visualizations for a comprehensive understanding.