Project Title: Time Series Analysis of India's GDP for Investment Insights
This repository documents a comprehensive time series analysis focused on forecasting India's GDP to evaluate the nation as a potential long-term investment opportunity. Throughout this project, we utilized a robust dataset sourced from the International Monetary Fund (IMF), covering annual GDP data from 1980 to 2023.
Key Skills and Tools Used:
Exploratory Data Analysis (EDA): Initial data investigation to understand trends and patterns.
Time-Series Decomposition: Applied linear, quadratic, cubic, and cyclical models to identify the model with the highest R², capturing both signal and noise components.
Model Selection and Fitting:
ARIMA Model: Developed an ARIMA model, tuning parameters based on the Durbin-Watson statistic for detecting autocorrelation.
Neural Networks: Employed neural networks for more sophisticated pattern recognition and forecasting, assessing model accuracy with Mean Absolute Percentage Error (MAPE) metrics.
Statistical Analysis: Used statistical tests like the Durbin-Watson to check for autocorrelation and adjusted models accordingly.
Forecasting and Business Insights: Generated forecasts for future GDP values, providing actionable insights for long-term investment strategies.
The repository includes R scripts, a detailed PowerPoint presentation of the project findings, and documentation on methodologies employed for model selection and forecasting accuracy assessment.