PROJECT PRELIMINARY REPORT

ON

“CRYPTOCURRENCY PRICE PREDICTION”

Submitted for the degree of

Bachelors of Information
Technology Engineering B.E.

ALIAKBAR KANOREWALA
YASH DUDHAWADE
SHIVAM SINGH

UNDER THE GUIDANCE OF

MRS MUDRA DOSHI

DEPARTMENT OF INFORMATION TECHNOLOGY

UNIVERSAL COLLEGE OF ENGINEERING

VASAI – 401 208.

UNIVERSITY OF MUMBAI
2023-2024
 PRELIMINARY REPORT APPROVAL FOR B.E.

This preliminary project report entitled

“CRYPTOCURRRENCY PRICE PREDICTION”
By
ALIAKBAR KANOREWALA
YASH DUDHAWADE
SHIVAM SINGH

is approved for the degree of Bachelors of Information Technology,

Mrs Mudra Doshi

[PROJECT GUIDE]

Mrs Mudra Doshi

[INTERNAL EXAMINER] [EXTERNAL EXAMINER]

Dr. Yogita Mane Dr. Jitendra Patil

[HEAD OF DEPARTMENT] [PRINCIPAL]

Place: Date :

i
 CERTIFICATE

This is to certify that the Project Report entitled “CRYPTOCURRENCY PRICE

PREDICTION” which is submitted by project members in fulfillment of the requirement
for the award of degree B.E in Information Technology to Mumbai University is a record of
the candidates own work carried out by them under my supervision. The matter embodied in
this thesis is original and has not been submitted for the award of any other degree.

Date:

Name of the Students:

YASH DUDHAWADE
ALIAKBAR KANOREWALA
SHIVAM SINGH

Name of the Guide: MRS MUDRA DOSHI

Sign and Date:

DR. YOGITA D. MANE DR. J. B. PATIL

HOD Principal
Department of I.T. UCOE, Kaman

MRS. MUDRA DOSHI DR/MR/MRS.

(Internal Examiner) (External Examine

i
 ACKNOWLEDGEMENT

This is a great pleasure and immense satisfaction to express our deepest sense of gratitude and
thanks to everyone who has helped us in completing our work successfully. We are presenting this
Project report on “CRYPTOCURRENCY PRICE PREDICTION” as part of the curriculum of B.E
Information Technology. Inspiration and guidance are invaluable in every aspect of life,
especially in the field of academics, which we have received from our respected
Project Guide: Mrs.Mudra Doshi and Head of Information Technology Department: Dr.
Yogita Mane. Besides, we take this opportunity to express our sincere gratitude to the
Principal: Dr. Jitendra Patil, UCOE for providing a good environment and facilities to complete
this project. We would also like to thank all our staff and friends who have directly or indirectly
guided and helped us in the preparation of this report and also for giving us an unending support
right from the stage this idea was conceived.

ALIAKBAR KANOREWALA
YASH DUDHAWADE
SHIVAM SINGH

iii
 ABSTRACT

In the past eight years of Bitcoin’s history, the economy has seen the price of Bitcoin rapidly grow due
to its promising outlook on the future for crypto currencies. Investors have taken note of several
advantages Bitcoin provides over the traditional banking system. One such trait is that Bitcoin allows
for decentralized banking, meaning that Bitcoin cannot be regulated by powerful banks. There is also a
market cap of 21 million Bitcoins that can be in circulation, therefore a surplus of Bitcoins cannot be
“printed” which would result in inflation. Bitcoin resolves the issue with transaction security by using
a block chain, or a ledger, which records the history of every transaction ever made into one long
hexadecimal “chain” of anonymous transactions, which keeps transaction history transparent, but also
confidential. Bitcoin as a result has become a very bullish investing opportunity, and due to the huge
volatility of the Bitcoin market price, this paper attempts to aid in investment decision making by
providing Bitcoin market price prediction. Our team explored several Machine Learning algorithms by
using Supervised Learning to train a prediction model and provide informative analysis of future market
prices. We start with Linear Regression models, and train on several important features, then We
proceed with the implementation of Recurrent Neural Networks (RNN) with Long Short Term Memory
(LSTM) cells. All code is written in Python using Google’s Tensor Flow software library. We show
that the price of Bitcoin can be predicted with Machine Learning with high degree of accuracy.

Keywords: Bitcoins, Crypto Currency, Deep Learning, Machine Learning

iii
CONTENT

Certificate …………………………………………………………………… I

Acknowledge ……...……………………………………………………………. Ii

Abstract …………………………………………………………………… Iii

List of Figures ……...……………………………………………………………. Vi

List of Tables …………………………………………………………………… Vi

Abbreviations

Introduction……………………………………………………………………. 01

Aims and Objective……………………………………………………………… 02

1.1 Aim ……………………………………………………………………………. 02

1.2 Objective ………………………………………………………........................ 02

iv
3. Literature Survey…………………………………………………….…. 03

3.1 Comparative Study.…………………………………………….. 05

4. Existing System…………………………………………………….…… 06

5. Problem Statement………………………………………………….… 07

6. Scope………………………………………………………………….…. 08

7. Proposed System…………………………………………………….…. 09

7.1 System Architecture……………………………………………. 10

7.2 Algorithm…………………………………………………….…. 10

8. Methodology………………………………………………………….…. 11

8.1 DFD Diagram…………………………………………………… 11

8.2 Sequence Diagram……………………………………………… 12

8.3 Use Case Diagram……………………………………………… 13

8.4 Activity Diagram………………………………………………. 14

8.5 Flowchart……………………………………………………….. 15

9. Planning ………………………………………………………….….…. 16

10. Details of Hardware and Software …………………………………… 17

11. Advantages ………………………………………………………….… 18

12. Implementation for next Semester …………………………………… 19

References ………………………………………………………….…… 20

v
 LIST OF FIGURES

Fig No. Name of Figure Page No.

4.1 Existing system structure 06

7.1 System Architecture 10

7.2.1 Algorithm 10

8.1.1 Level 0 DFD 11

8.1.2 Level 1 DFD 11

8.2 Sequence Diagram 12

8.3 Use Case Diagram 13

8.4 Activity Diagram 14

8.5 Flowchart 15

9.1 Gantt Chart 16

vi
LIST OF TABLES

Table No. Table Name Page No.

3.1 Comparative Analysis Existing System………………… 0

vii
ABBREVIATIONS

Abbreviations Description

D-APP Decentralized application

ETH Ethereum

NFT Non Fungible Task

IPFS Inter Planetary File system

DFD Data Flow Diagram

CLI Command Line Interface

viii
1.INTRODUCTION

In the ever-evolving realm of finance, cryptocurrencies have transcended their origins to become a global
phenomenon, reshaping traditional concepts of investment and finance. At the forefront of this digital
revolution stand a handful of cryptocurrencies—Bitcoin, Ethereum, Dogecoin, Shiba Inu, and Meta—
each representing a unique facet of this decentralized landscape. The surge in interest and investment
within the cryptocurrency market has spurred an unprecedented need for reliable insights and predictive
tools. Understanding the complexities and volatility inherent in this space, this project seeks to pioneer
an innovative solution—a mobile application dedicated to forecasting the price movements of these five
influential cryptocurrencies. The exponential growth and unpredictability of cryptocurrency markets
necessitate a multifaceted approach.

Thus, this application endeavors to amalgamate cutting-edge technologies with robust predictive
methodologies to offer users a comprehensive and insightful platform. Beyond mere speculation, this app
aims to distill intricate market data into actionable insights through a fusion of predictive models. These
models, incorporating technical analysis, fundamental indicators, sentiment analysis from diverse
sources, and adaptive machine learning algorithms, aim to unveil potential future price trajectories. Yet,
this project is not solely about predictive analytics; it is equally dedicated to educating and empowering
users. Through intuitive interfaces and real-time updates, the app aims to demystify the complexities of
cryptocurrency markets.

By providing accessible explanations of market indicators and fostering a space for collaborative
learning, the app seeks to level the playing field for both seasoned investors and newcomers. The
significance of this endeavor lies not just in predicting prices but in fostering a community-driven
platform that equips users with the knowledge and tools necessary to navigate the volatile yet promising
landscape of cryptocurrencies. By harnessing the power of technology and data-driven insights, this app
aspires to empower users in making informed decisions in their cryptocurrency investment endeavors.
This project stands as a testament to the fusion of technology, finance, and education—a bridge between
intricate market analysis and accessible information, aimed at shaping a more informed and empowered
community within the captivating domain of cryptocurrencies.

6
2.AIM AND OBJECTIVE

2.1 Aim

The aim of this project is to create a web application that achieves cryptocurrency price prediction
of crypto by analyzing the past historical data using ML models. Develop a Reliable Predictive
Tool which gives the best accuracy.

2.2 Objectives
:

● Acquire diverse datasets from cryptocurrency exchanges, financial APIs, social

media platforms, and news sources.
● Conduct thorough data preprocessing to handle missing values, outliers, and
inconsistencies.
● Implement various machine learning algorithms, including regression,
classification, and time-series forecasting models.
● Train models on historical cryptocurrency data to learn patterns and relationships.
● Optimize model hyperparameters using techniques like grid search, random search,
and Bayesian optimization.
● Validate model robustness through cross-validation and bootstrapping methods.

● Develop a scalable and responsive real-time prediction platform using Flutter for
cross-platform compatibility.
● Integrate APIs for fetching live cryptocurrency data and model predictions

6
 3. LITERATURE SURVEY

Author Project Title Publication Purpose

Makridakis et al. Cryptocurrency Forecasting with Journal of Risk and To evaluate the
Neural Networks for Decision Financial Management performance of neural
Making network models in
cryptocurrency price
prediction
Singh et al. A Hybrid AI Model for IEEE Access To propose a hybrid model
Cryptocurrency Price Prediction combining LSTM and
Using LSTM and ARIMA ARIMA for improved
prediction accuracy
Garcia et al. Do Cryptocurrencies Follow Achieved enhanced To investigate the
Financial Market Fractals? accuracy compared to presence of fractal patterns
Evidence from a Copula-Based statistical methods. in cryptocurrency markets
Analysis using copula analysis
Kim et al. Use of ANN and RF for stock Journal of Risk and To assess the effectiveness
closing price prediction. Financial Management of deep learning
techniques, such as CNNs
and LSTMs, in
cryptocurrency price
prediction
Lee et al. Predicting Cryptocurrency Prices Journal of Risk and To explore the impact of
Using Deep Learning Techniques Financial Management sentiment analysis on
cryptocurrency price
prediction

Zhang et al. Cryptocurrency Price Prediction Expert Systems with To investigate the
Using Sentiment Analysis Applications correlation between
Twitter sentiment and
Bitcoin price fluctuations

6
4. EXISTING SYSTEM

The existing systems for cryptocurrency price prediction predominantly rely on historical data
analysis, technical indicators, and statistical models. These systems often employ methodologies
like moving averages, relative strength index (RSI), and other technical analysis tools to forecast
future price movements. Additionally, machine learning models, including regression, decision
trees, and support vector machines (SVM), are utilized to capture patterns and trends within
historical price data. However, these systems often face challenges in handling the extreme

6
5. PROBLEM STATEMENT

In the dynamic landscape of cryptocurrency markets, the accurate prediction of price movements remains
a challenge due to the inherent volatility and complexity of these digital assets. Existing systems primarily
reliant on historical data analysis and technical indicators often encounter limitations in capturing the
real-time sentiments and swiftly adapting to rapid market shifts. The absence of robust methodologies
that effectively integrate real-time emotional insights, emerging trends, and fundamental indicators
hampers the precision and reliability of cryptocurrency price predictions.

As a result, investors and traders face uncertainty and risk in decision-making, lacking comprehensive
tools that holistically consider the multifaceted factors influencing cryptocurrency prices. Addressing
these challenges necessitates the development of advanced prediction models capable of assimilating
real-time emotional and market sentiments, thereby providing more accurate and timely insights for
effective investment strategies in the volatile cryptocurrency landscape.

13
6. SCOPE

Integrating Advanced Analysis in Cryptocurrency Price

Prediction:
The fusion of advanced analytical models with cryptocurrency price prediction holds expansive
potential across various domains and applications within the financial landscape.

Financial Services and Investment Strategies:

In the realm of financial services, integrating cutting-edge predictive

models into cryptocurrency price analysis could revolutionize

investment strategies.

Risk Management and Decision Support:

Advanced predictive analytics in cryptocurrency markets can

significantly benefit risk management strategies. Real-time analysis of

emotional indicators and market sentiments could provide traders and

investors with timely signals.

Algorithmic Trading and Market Efficiency:

The integration of emotion-aware predictive models could also pave the way for more efficient

algorithmic trading systems in the cryptocurrency space. By assimilating

emotional cues and real-time

13
 7.1 SYSTEM ARCHITECTURE

Fig 7.1 System Architecture

The Cryptocurrency Price Prediction Platform's architecture is structured for efficient data collection,
processing, and analysis. It integrates data from multiple sources, applies machine learning algorithms
for price prediction, and delivers real-time insights through a user-friendly interface. Data undergoes
preprocessing to ensure accuracy, while trained models generate forecasts based on historical trends. The
platform's user interface offers intuitive navigation and customization options. Security measures
safeguard sensitive data, and scalability features ensure performance as user demand grows. With a focus
on reliability and accuracy, the platform empowers users to make informed decisions in the dynamic
cryptocurrency market landscape.
13
 8.METHODOLOGY

8.1 Level 0 DFD

Fig 8.1 Level 0 DFD

The level 0 data flow diagram (DFD) for the Cryptocurrency Price Prediction Platform depicts interactions between external entities (users
and data sources) and internal processes (Prediction Engine, Data Collection and Processing, and User Interface). Data flows from external
sources to the platform for ingestion and preprocessing, then to the Prediction Engine for analysis and prediction generation. Results are
displayed to users via the User Interface process. User inputs flow back into the system for processing.

13
Level 1 DFD

Fig 8.2 Level 1 DFD

At level 1 of the Cryptocurrency Price Prediction Platform's data flow diagram, external data sources provide raw data. This data is
collected and preprocessed to ensure accuracy and usability. Relevant features are engineered, and machine learning models are trained
using historical data. These models generate real-time predictions, which are displayed to users through the user interface. User
feedback is managed to improve system performance. This breakdown highlights the flow of data and processes within the platform,
from data collection to user interaction and feedback incorporation, enabling accurate cryptocurrency price predictions and informed
decision-making.

13
8.2 Sequence Diagram

Fig 8.2 Sequence Diagram

This sequence diagram illustrates the process of cryptocurrency price prediction. It begins with the user requesting
data, followed by the prediction model receiving, processing, and generating a prediction. Finally, the prediction
is sent to the user interface, which displays it to the user.

13
8.3 Use Case Diagram

Cryptocurrency Price Prediction System

User

Cryptocurrency Price Prediction System

Collect Historical Data

Process Data

Train Prediction Model

Evaluation Model

Deploy Model for Real-Time Cryptocurrency

Price Prediction

Display Prediction Result to User

Fig 8.3 Use Case Diagram

The use case diagram depicts interactions within a cryptocurrency price prediction system. It includes actors such as the user and the system
itself, detailing processes like data collection, processing, training, model evaluation, deployment, and displaying prediction results to users.

13
8.4 Activity Diagram

Cryptocurrency Price Prediction Hyperparameter Tuning

write

Model Deployment
Start Process
(Integration with prediction
System)

Data collection
(historical Data)
Real time prediction
(New Data Input)

Data Preprocessing
(Cleaning, Scaling, Feature
Engineering) Display Results
(Charts, Alerts)

Model Training
End Process
(Machine Learning/Deep Learning)

Model Evaluation
(RME, RMSE, MSE, MAE)

Fig 8.4 Activity Diagram

The activity diagram illustrates the steps involved in cryptocurrency price prediction, including data collection,
preprocessing, model training, hyperparameter tuning, deployment, real-time prediction, and displaying results
14
8.5 Flow Chart

Fig 8.5 Flow Chart

This flowchart illustrates the steps in real-time cryptocurrency price prediction using deep learning. It begins with data collection,
preprocessing, model training, evaluation, and hyperparameter tuning. Then, the model is deployed for real-time prediction, and results are
displayed to users.

15
9. PLANNING

Fig. 9.1 Gantt chart for planning

16
10. DETAILS HARDWARE AND SOFTWARE

10.1 Hardware Requirements

System : Intel Core i3 2.00 GHz.

Hard Disk : 1 TB.

Monitor : 14’ Color Monitor.

Mouse : Optical Mouse.

Ram : 2 GB.

Keyboard : 101 Keyboard Keys.

10.2 Software Requirements

Operating system : Windows 10.

Coding Language : Flutter, jupyter notebook

Software’s used : Android studio

17
11. ADVANTAGES

• Personalized Investment Insights: Integrating sentiment analysis and emotional recognition into
cryptocurrency price prediction systems can personalize insights for investors. By analyzing emotional
indicators, these systems could offer tailored investment recommendations aligned with an investor's risk
tolerance, potentially enhancing their decision-making process.

• Increased User Engagement: Emotion-integrated prediction models in cryptocurrency analysis might

enhance user engagement by providing emotionally resonant insights. Investors may find themselves more
connected to market data when it's presented with emotional context, potentially leading to sustained
interest and active participation.

• Decision Support for Risk Management: Emotion-aware cryptocurrency prediction systems could assist
in managing investment risks by factoring in emotional sentiment analysis. These systems might offer
alerts or insights based on emotional indicators, aiding investors in making more informed decisions and
potentially mitigating risks associated with impulsive trading.

• Mental Well-being and Stress Reduction: The integration of emotional recognition within cryptocurrency
analytics might indirectly support investors' mental well-being. Providing insights or recommendations
aligned with market sentiments could help reduce stress related to investment decisions, contributing to a
more composed approach to trading.

• Inclusivity and Accessibility: Similar to its application in music technology, emotion recognition in
cryptocurrency analytics could make investment insights more accessible. It could cater to investors who
might find traditional market analysis overwhelming, providing a more user-friendly and inclusive
approach to understanding market trends.

20
12. IMPLEMENTATION FOR NEXT SEMESTER

In the later stage of the project we will try to improve the web app's user interface and the user
experience for the users. We will try to test it under different conditions and incorporate new
features. A major focus will be there to improve the accuracy of the model

20
REFERENCES
1 Bitcoin Price Index - Real-time Bitcoin Price Charts", CoinDesk Available:
https://www.coindesk.com/price

2. A. Ng, "Linear Regression With Multi Variable", Stanford, CA

3. D. Nelson, A. Pereira and R. de Oliveira, "Stock Market’s Price Movement Prediction With LSTM
Neural Networks", in Neural Networks (IJCNN).

4. M. Dixon, D. Klabjan and J. Bang, "Classification-based Financial Markets Prediction using Deep
Neural Networks", Illinois Institute of Technology. https://doi.org/10.1162/neco.1997.9.8.1735

5. S. Hochreiter and J. Schmidhuber, “Long Short-Term Memory,” Neural Computation, vol. 9, no.
8, pp. 1735–1780.

6. D. Shah and K. Zhang, "Bayesian regression and Bitcoin", Massachusetts Institute of Technology.

7. I.Georgoula, D. Pournarakis, C. Bilanakos, D. Sotiropoulos and G. Giaglis, "Using Time-Series

and Sentiment Analysis to Detect the Determinants of Bitcoin Prices

20