Download as pdf or txt
Download as pdf or txt
You are on page 1of 8

Machine Learning Mini Project Report on

Bitcoin Price Prediction

By

CHAITANY PRAKASH GHADIGAONKAR (ROLL NO.: 10)


SURAJ NANDKUMAR SAKHARE (ROLL NO.: 11)
TAJ MOHAMMAD BADRUDUJJA KHAN (ROLL NO.: 12)

Guide:
PROF. JANHAVI SANGOI

Department of Computer Engineering


VIVA Institute of Technology

University of Mumbai
2023 - 24
1.0 Abstract
____________________________________________________
The rapid evolution of cryptocurrencies, most notably Bitcoin, has garnered significant attention
in the financial and technological domains. Predicting the price movements of Bitcoin is of
paramount interest for traders, investors, and financial institutions. Machine learning techniques
have emerged as a powerful tool for forecasting asset prices in volatile markets, and this project
focuses on utilizing these methodologies to predict Bitcoin price fluctuations.
This project's primary objective is to develop and evaluate machine learning models for Bitcoin
price prediction. We will employ a diverse dataset that includes historical Bitcoin price data,
market sentiment indicators, trading volumes, and other relevant features. The dataset spans a
substantial time frame to capture various market conditions, including periods of high volatility
and stability. The project will explore different machine learning algorithms, such as regression,
time series forecasting, and deep learning models, to identify the most effective approach for
Bitcoin price prediction.
The outcomes of this project will contribute to a better understanding of Bitcoin price dynamics
and may offer valuable insights to investors and traders looking to make informed decisions in
the cryptocurrency market. Moreover, the project's findings could serve as a foundation for the
development of trading strategies and risk management techniques in the evolving world of
digital assets.
2.0 Introduction
____________________________________________________
The emergence of Bitcoin, a decentralized digital currency, has revolutionized the financial
landscape, challenging traditional monetary systems and paving the way for a new era of digital
finance. As Bitcoin has gained widespread adoption and recognition, its price dynamics have
become increasingly complex, marked by extreme volatility and significant fluctuations. The
ability to accurately predict these price movements is of paramount interest to investors, traders,
and financial institutions in the cryptocurrency space. Machine learning, a subfield of artificial
intelligence, has proven to be a powerful tool for analyzing and forecasting financial markets.
This project endeavors to harness the potential of machine learning to predict Bitcoin price
changes, offering a novel approach to understanding and navigating the dynamic cryptocurrency
market. Bitcoin's price behavior is influenced by a multitude of factors, including market
sentiment, trading volumes, macroeconomic events, and the interplay between supply and
demand. In this project, we aim to leverage a wide range of data sources, encompassing
historical price data, market indicators. By incorporating these diverse data streams, we seek to
build predictive models that can capture the intricate patterns and relationships within the Bitcoin
market, ultimately enabling more informed decision-making. The cryptocurrency market is
unique in its characteristics, often exhibiting non-linear and erratic price movements. As a result,
conventional financial models may not be suitable for effectively forecasting Bitcoin's price.
3.0 Project Scope
____________________________________________________
The scope of this machine learning project for Bitcoin price prediction is multifaceted and
comprehensive. It begins with data collection, encompassing historical Bitcoin price data, market
indicators, and textual data for sentiment analysis, ensuring data quality and consistency
throughout the project. Data preprocessing includes cleaning, handling missing values, and
standardizing features. Feature engineering explores the creation of new features and
transformations to enhance model performance, including time-based features, technical
indicators, and sentiment scores. Various machine learning models, ranging from linear
regression to deep learning, will be considered, and model selection will depend on their
performance. Hyperparameter tuning will fine-tune model parameters using cross-validation
techniques. Sentiment analysis will be integrated to analyze the impact of market sentiment on
predictions. Model evaluation will use metrics like MAE and RMSE, with backtesting to assess
trading strategies. The project will provide interpretability through visualizations and feature
importance analysis, comparing the performance of different models. Documentation and
reporting will be comprehensive, offering insights and recommendations. Future research
possibilities will be highlighted, and the project timeline will be well-defined, accounting for all
project phases. Stakeholder collaboration will ensure alignment with specific needs and
incorporate valuable feedback into the project.
4.0 Project Description and Implementation
___________________________________________________
Project Description:
This project seeks to create a robust and data-driven Bitcoin Price Prediction system using
machine learning methodologies. The dynamic cryptocurrency market has made accurate price
forecasting of Bitcoin, the most prominent digital asset, a critical endeavor for investors and
institutions. By gathering and preprocessing historical Bitcoin price data from trusted sources,
the project will lay the foundation for reliable predictions. The incorporation of feature
engineering, including technical indicators and sentiment analysis, aims to enhance the models
predictive capabilities. A variety of machine learning models, ranging from classical regression
to advanced deep learning, will be explored and evaluated for their effectiveness in capturing
Bitcoin price trends. The inclusion of sentiment analysis will enable the project to analyze the
influence of market sentiment on Bitcoin's price dynamics. Model evaluation will employ a set
of metrics to assess performance, including Mean Absolute Error and Root Mean Square Error.
The project will also produce visualizations and provide comprehensive documentation, offering
valuable insights for investors, traders, and financial institutions navigating the cryptocurrency
space. Ultimately, this project will serve as a pivotal exploration of Bitcoin price prediction,
harnessing the power of machine learning and data analysis to navigate the complexities of
cryptocurrency markets.

Hardware and software requirements:


Hardware requirements:
● RAM: minimum 4GB ram.
● ROM: minimum storage space required to store the data
Software requirements:
● Operating system: Windows, MacOS, Linux
● IDE: Jupyter notebook, pycharm
● Machine learning libraries: Numpy, Pandas, Scikit-Learn
● Data visualization libraries: Matplotlib
● Text analysis libraries: NLTK
Implementation :

Fig 4.1: Block Diagram


The figure 4.1 represents the block diagram of bitcoin price prediction.

Code:
# Load and preprocess your time series data
# X_train, y_train, X_test = ...

# Define and compile the LSTM model


model = Sequential()
model.add(LSTM(units=50, activation='relu', input_shape=(X_train.shape[1], X_train.shape[2]))
model.add(Dense(1))
model.compile(optimizer='adam', loss='mean_squared_error')

# Train the model


model.fit(X_train, y_train, epochs=50, batch_size=32)

# Make predictions
predictions = model.predict(X_test)
5.0 Outputs of Project
____________________________________________________

Fig 5.1: Next 30 days predicted price


The figure 5.1 shows the next 30 days predicted price of the bitcoin based on the previous data.
The figure shows the last 15 days data and next 30 days predicted data.

Fig 5.2: Close price prediction


The figure 5.2 shows the predicted close price of the bitcoin. The model uses the last data and
predict the next close price of the bitcoin data.
6.0 Conclusion
____________________________________________________
Bitcoin price prediction project, with its machine learning model offers a valuable tool for
investors and cryptocurrency enthusiasts. By leveraging historical data, technical indicators, and
sentiment analysis, the model provides informed predictions, aiding in decision-making and
trading strategies. The integration of real-time data updates ensures relevance in an ever-
evolving market. While this project simplifies Bitcoin price forecasting, it also serves as a
stepping stone for future enhancements and research opportunities. In a rapidly changing
landscape, this application contributes to the empowerment of users, promoting data-driven
decisions in the cryptocurrency market.

You might also like