Skip to content

The analysis reveals the challenges of predicting Bitcoin prices during highly volatile periods and demonstrates how traditional time series models perform under different market conditions. The project includes comparative analysis of model performance during stable and volatile market phases.

Notifications You must be signed in to change notification settings

Urvee1810/Bitcoin-price-forecasting-using-ARMA

Repository files navigation

Bitcoin-price-forecasting-using-ARMA

This project was completed as part of PG Level Advanced Certification Programme in Computational Data Science coursework at Centre for Continuing Education - Indian Institute of Science in collaboration with Talent Sprint

A special thanks to Prof. Dr. Shashi Jain & Mentor Mr. Sachin Sharma

Problem Statement: Perform EDA and forecast the Bitcoin price using ARMA model on timeseries (bitcoin) data.

Module: Business Analytics

Project Type: Team

Project Objective

  • perform EDA on time series data
  • analyze the auto correlation and partial auto correlation plots
  • implement the ARMA model and forecast the bit coin price Here's a clear, descriptive summary for your GitHub repository:

Overview

A comprehensive analysis of Bitcoin price movements using time series analysis techniques. This project explores historical Bitcoin price data from September 2014 to July 2021, implementing various technical indicators and forecasting models.

Key Features

  • Technical indicator implementation (Bollinger Bands, MACD, Stochastic Oscillator)
  • Time series analysis with ARIMA modeling
  • Feature importance analysis using Random Forest
  • Stationarity testing and seasonal decomposition
  • Price prediction with confidence intervals

Technical Components

  • Time series decomposition and stationarity analysis
  • ARMA model parameter optimization
  • Volatility impact analysis on model performance
  • Interactive visualizations using mplfinance

Tools & Technologies

  • Python
  • Pandas & NumPy
  • Scikit-learn
  • Statsmodels
  • Matplotlib & Seaborn
  • mplfinance

Results

The analysis reveals the challenges of predicting Bitcoin prices during highly volatile periods and demonstrates how traditional time series models perform under different market conditions. The project includes comparative analysis of model performance during stable and volatile market phases.

Usage

The Jupyter notebook contains detailed analysis and code implementation, including:

  • Data preprocessing
  • Technical indicator calculation
  • Model training and evaluation
  • Visualization of results

About

The analysis reveals the challenges of predicting Bitcoin prices during highly volatile periods and demonstrates how traditional time series models perform under different market conditions. The project includes comparative analysis of model performance during stable and volatile market phases.

Topics

Resources

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published