Bitcoin Price Prediction Using Machine

Learning Algorithms

P. Nagamani(B) , G. Jaya Anand, S. Ganga Prasanna, B. Sai Raju,

and M. H. S. V. Siva Satish

Computer Science and Engineering, Godavari Institute of Engineering and Technology

(Autonomous), Rajamahendravaram, India
nagamani@[Link], nagamanipedapati@[Link]

Abstract. The past several years have seen an increase in interest in trading
that is supported by machine learning and artificial intelligence. Utilize auto-
mated trading with the aid of machine learning and artificial intelligence to reap
the maximum rewards from the cryptocurrency market. For a specific time, we
keep the daily data. We achieve excellent results by utilising tactics supported by
cutting-edge algorithms. The results produced the expansion in the crypto cur-
rency industry with the aid of straight forward architecture and algorithms. The
rise in market capitalization has led to a rise in popularity for the cryptocurrency
in 2017. Today’s market involves more than 1500 crypto currencies. For usage
in online transactions, the crypto currency can be created. A crypto money tech-
nology is bitcoin. Bitcoin’s value changes constantly, second by second. As a
result, we apply machine learning architecture to forecast the value of the bitcoin
price in this case. We are working to demonstrate that, in comparison to previ-
ous techniques and architectures, this ML architecture produces results that are
more accurate. Our study use the Support Vector Machine(SVM) and K Nearest
Neighbor(KNN)algorithms to successfully forecast bitcoin prices. The findings
demonstrate that the Support Vector Machine(SVM) method outperforms the K
Nearest Neighbor(KNN) method as it is currently being used.

Keyword: Bitcoi · machinelearning · KNN · SVM · dataset · classification ·

regression · prediction · accuracy

1 Introduction

Bitcoin and other digital currencies are commonly used for speculative or advanced
payment purposes. For instance, Bitcoin is decentralised because no one owns it. Because
they are not limited to a single nation, bitcoin exchanges are simple. Through a number
of commercial venues referred to as “bitcoin trades,” speculation ought to be accessible.
These allow users to trade bitcoins for a wide range of foreign currencies. The most
Bitcoin is traded on Mt. Gox. Bitcoins are stored in a smart wallet that works like a
virtual bank account. Block chain is a database that keeps track of various transactions’
dates and records. In a block chain, a square stands in for each record. Each informational

© The Author(s) 2023

B. Raj et al. (Eds.): ICETE 2023, AER 223, pp. 389–396, 2023.
[Link]
390 P. Nagamani et al.

square includes a link to the preceding tile. Data on the blockchain is encrypted. During
exchanges, only the customer’s wallet ID is made visible; their name is not. Similar to
a stock, Bitcoin’s value fluctuates, but in a strange way. Digital currencies like Bitcoin
are used all around the world for speculation and advance payments.
For instance, Bitcoin is decentralised because no one owns it. Given that they are
not bound to any certain nation, bitcoin exchanges are simple. Through several business
hubs known as “bitcoin exchanges,” speculation should be possible. These allow users
to trade bitcoins for a wide range of foreign currencies. The most Bitcoin is traded on
Mt. Gox. A smart wallet, which functions like a virtual bank account, is where bitcoins
are kept. The block chain is the repository for the timestamp data and the records of a
significant number of deals. In a block chain, each record is represented by a square.
Every informational square contains a reference to a previous tile. On the blockchain,
data is encrypted. Only the wallet ID of the customer is made visible during exchanges;
their name is not.
Like a stock, Bitcoin’s value changes, but in an unexpected way. Using information
from financial trades and a range of calculations, value is forecasted. But Bitcoin is
impacted by peculiar variables. Forecasting the value of Bitcoin is crucial for making the
best business decisions. The price of bitcoin is unaffected by governmental intermediaries
or changes in the economy, in contrast to a stock market. Therefore, we consider it crucial
to estimate the value of Bitcoin using cutting-edge AI technology.

2 Literature Survey
New technologies and methods have emerged as a result of the Internet’s exponential
access development. One way that the Internet is replacing the conventional monetary
system as a means of exchange is through the use of cryptocurrencies. Cryptocurrencies,
also known as digital or virtual currencies, are used to trade or transfer assets over a net-
work of computers. In a brief period of time, the market for cryptocurrencies has grown
exponentially. In 2009, the moniker “Bitcoin” was given to the first cryptocurrency.
Decision Tree and Regression Techniques for Predicting Cryptocurrency Prices
Karunya Rathan; SomarouthuVenkat Sai Tubati Sai Manikanta Publisher: IEEE 2021.
In the first part of the study, the price trend on daily variations in the price of bitcoin is
detected while providing information on price trends for bitcoin [1].
Using machine learning, predict the price of bitcoin Siddhi Velankar; Sakshi
ValechaShreya MajiPublisher: IEEE 2021. In this essay, we make an effort to fore-
cast the price of bitcoin with some degree of accuracy while taking into account many
factors that influence its value [2].
Deep Learning and Neural Networks for Prediction of Cryptocurrency Prices Toby
Biswas; Mohandas Pawar;IEEE [Link] digital currencies have such a broad price
range, it isessentialto have solid planning when attempting to estimate the priceof a given
currency [3]. Short-Term Cryptocurrency Price Movement Prediction Using Centrality
MeasuresKin-Hon Ho; Wai-Han Chiu;Chin Li Publisher:IEEE 2021. In this essay,we
make an effort to forecast the price of bitcoin with some degree of accuracy while taking
into account many factors that influence its value [4].
 Bitcoin Price Prediction Using Machine Learning Algorithms 391

Ether Price Prediction Using Advanced DeepLearning ModelsAgis Politis; Ka-

terinaDoka;Nectarios Koziris Publisher: IEEE 2021. Although several re-search ini-
tiatives have been made to create systems that can accurately fore-cast price fluctuations
in the bitcoin market, they show substantial efficiency gaps, which this study further
analyses [5].

3 Proposed Methodology

The proposed system includes the algorithms such as KNN and SVM that are used to
predict the prices of Bitcoins and giving best accurate results for the future value of
crypto currencies.
Even so, there are some unanticipated fluctuations in the value of the Bitcoin. On
financial exchange data, different calculations are used to predict [Link], Bit-
coin is affected by some unusual factors. This research aims to determine the pre-
cision with which machine learning techniques can forecast the direction of Bitcoin
[Link], this is a time series prediction issue. Although a tonne of study
has been done on the application of different machine learning techniques to predicr
time series, that hasn’t been much done in this area specifically on [Link],
because it is still in its infancy compared to other currencies like the USD, Bitcoin is
much more unpredictable. In order to predict the price of bitcoin based on the taken
into account and further described historical data, we used the SVM and KNN machine
learning algorithms.

3.1 Machine Learning Algorithms

A branch of computer science and artificial intelligence (AI), the study of machine learn-
ing seeks to pretend human learning processes through the use of data and algorithms,
gradually improving the results’ accuracy. Machine learning is a vital part of the develop-
ing field of data science. Data mining projects use algorithms that have been statistically
taught to make classifications or predictions. The decisions made as a result of these
insights affect key growth metrics and are then used to fuel applications and compa-
[Link] data’s continued growth and expansion will drive up demand for data scientists,
who will be required to assist in identifying the most crucial company issues, as well as
the data required to address them. Machine learning relies on input, such as training data
or knowledge graphs, in a similar way to how the human brain learns information and
develops understanding in order to grasp entities, domains, and the connections among
them. Deep learning cannot begin until entities are [Link] first step in machine
learning is observation or data, such as samples, first-hand knowledge, or instructions.
K-Nearest Neighbor:
The key details in remote sensing images can only be recovered once they have been
accurately classified. It is essential to use classification to take the fine information out
of an image for furtherprocessing. This study shows that active learning methods were
utilised to determine the best classifier in hyper spectral images and that KNN techniques
were evaluated in those images. The best classification method has recently been a hot
topic in academia. The K-nearest neighbourhood technique is frequently used to classify
392 P. Nagamani et al.

photographs. In order to combine the location using the greatest margin classification,
a better KNN is used for high resolution remote sensing. A synthetic immunological
B-cell network and KNN are used to demonstrate how processing can be done with less
data.
As an illustration, imagine that we have a picture of a creature that resembles both a
cat and a canine, but we are unsure of its identity. It is therefore possible to identify these
objects using the KNN technique, which is depending on a similarity metric. Our KNN
model will search for similarities between the traits of the new data set and those are the
images of cats and canines and, in accordance with those similarities, it will categorise
the new data set as cat- or dog-related.
Feature Selection.
By employing only pertinent data and obtaining a noise-free grid of input data. It is a
technique for decreasing the input variable in our model. It is the process of automatically
selecting pertinent characteristics for our machine learning model and solving every issue
according to the kind of machine learning algorithms you utilised in your model.
Support Vector Machine(SVM).
The SVM’s main objective is to search the appropriate hyperplane in a high-
dimensional space under various conditions. Numerous hyperplanes can realise this
paradigm. In this approach, the support vector is applied to the data that is closest to the
closed surface and the optimum choice surface. In order to accomplish classification, a
hyperplane is created to partition the data, and the input vectors are then planned into
a high dimensional space. Most non-convex, unconstrained minimization issues as well
as quadratic programming issues can be solved using this method. The most efficient
method for developing classifiers is the SVM (Fig. 1).

3.2 A Comparison of the Proposed and Existing System

In this work, the KNN and SVM algorithms are used to accurately predict the price
of bitcoin. The outcome demonstrated that the proposed SVM performs better than the
current KNN. The main flaw in the current system is its lack of accuracy, but with the
suggested method, prediction accuracy is good (Fig. 2).

4 Working Architecture of Proposed Methodology

4.1 Modules Description
1. Dataset Input.
2. Dataset Analysis.
3. Oversampling (using SMOTE).
4. Training and Testing Subset.
5. Using the algorithm.
6. Making Predictions about Outcomes.

1. Dataset Input:
You can get the dataset from an online data provider using online resources. In order
to accurately estimate the accuracy, we must amass a sizable collection of data.
 Bitcoin Price Prediction Using Machine Learning Algorithms 393

Fig. 1. Process of Proposed Model

Fig. 2. System Architecture

2. Dataset Analysis:
This Section Contains Dataset Analysis. For the Data Processing, the Data Size is
Taken into Account.
3. Oversampling (Using SMOTE):
394 P. Nagamani et al.

We Have Compiled a Thorough History of All day today’s Bitcoin Prices Over a
Significant Period of Time.
4. Training and Testing Subset:
Many Classifiers Exhibit Bias for Majority Classes Because the Dataset is Unbal-
anced. Minority-Class Characteristics Are Dismissed as Noise and Ignored. Therefore,
Choosing a Sample Dataset is Suggested.
5. Using the Algorithm:
The Classification Algorithms that Were Tested on the Dataset for the Sub-Sample
Are Listed Below. Knn and Svm Are Two Examples.
6. Making Predictions About Outcomes:
The Training Model is Used with the Test [Link] is the Metric that is
Utilised. The Desired Outcomes Are Obtained Once the ROC Curve is Displayed.

5 Result

The Fig. 3 shows the Bitcoin price is predicted by algorithms namely KNN and SVM
in terms of [Link] the result its proved that proposed system SVM works better
than existing KNN.
Here the Fig. 4 shows how the machine learning algorithms (KNN & SVM) gives
an accurate results for the future value of Bitcoin in 30 days.
The minimal price of Bitcoin projections for the forthcoming days are provided
Fig. 5, and the actual value we anticipated for that specific days is also included in the
below figure. This is based on an analysis of the values of Bitcoin in previous years. The
Fig. 5 only applies to our dataset because it is based on data from prior years.

Fig. 3. Accuracy Comparison of SVM & KNN

 Bitcoin Price Prediction Using Machine Learning Algorithms 395

Fig. 4. Bitcoins Price History

Fig. 5. Prediction Value

6 Conclusion

The finding of this investigation lead us to the following conclusion:

Recently, we integrated a different types of machine learning modules for the purpose
of predicting bitcoin price. Find out how accurate SVM and KNN are looking at the
comparision table of ML algorithm model corretness. SVM and KNN are shows to be
the most efficient algorithms in terms of reducing time complexity when using the ML
algorithm to predict bitcoin price. The techniques in machine learning will improve the
idea of a cryptocurrency feature. As a result, investments in globules will appreciate in
value. In this study, we developed a fresh method to judge the accuracy of the feature
pricing. This boosts revenue and customer expansion.

References
1. Fusion in cryptocurrency price prediction: a decade review on recent breakthroughs, archi-
tecture, and possible future orientations. Nisarg P. Patel et al. IEEE Access 10: 34511–34538
(2022).
2. “Performance evaluation of deep learning and boosted trees for bitcoin closing price pre-
diction,” by Azeez A. Oyedele et al. 2023: 119233 Expert Systems with Applications
213
3. Dolatsara, Hamidreza Ahady, et al. “An interpretable decision-support systems for daily
cryptocurrency trading.” Expert Systems with Applications 203 (2022): 117409.i
4. Making Bitcoin Price Predictions Using Machine Learning Ireland, Dublin, IEEE 2018, Sean
McNally, Jason Roche, and Simon Caton
5. Erfanian, Sahar, et al. “Predicting Bitcoin (BTC) Price in the Context of Economic Theories:
A Machine Learning Approach.” Entropy 24.10 (2022): 1487.
396 P. Nagamani et al.

6. Bi-LSTM Network used to predict the price of bitcoin. International Conference on Infor-
matics and Computer Communication in 2021. P. Nithyakani and others (ICCCI). IEEE,
2021.
7. “Optimizing resource allocation with intelligent agents,” by Lucas O. Souza, Celia G. Ralha,
and Bruno W.P. Hoelz. 2017’s 16th Conference on Autonomous Agents and Multi-Agent
Systems Proceedings.
8. Jacob Coburn and Sara C. Pryor’s “Projecting Future Energy Production from Operating Wind
Farms in North America: Part II: Statistical Downscaling.” Journal of Applied Meteorology
and Climatology 62.1; 2023: 81–101.
9. Gupta, Shresth, et al. “Higher Order Derivative-Based Integrated Model for Cuff-Less Blood
Pressure Estimation and Stratification Using PPG Signals.” IEEE Sensors Journal 22.22
(2022): 22030–22039.
10. “Spatial prediction of groundwater potential and driving factor analysis based on deep learning
and geographical detector in a dry endorheic basin,” by Wang, Zitao, Jianping Wang, and
Jinjun Han 109256 is the ecological indicator number for 2022.
11. “Holistic Approaches to Music Genre Classification Utilizing Efficient Transfer and Deep
Learning Methods,” by Prabhakar, Sunil Kumar, and Seong-Whan Lee. 211 (2023): 118636
in Expert Systems with Applications.
12. Chandra, MunipalliSasi, R. Sumathi, and J. Jeyaranjani. “Analysis of Predicting Bitcoin Price
using Deep Learning Technique.” 2022 3rd International Conference on Smart Electronics
and Communication (ICOSEC). IEEE, 2022.
13. The trio of Ye, Xuan, and Huang. Bayesian LASSO Analysis Using High Dimensional Data
BDCPS 2019, Shenyang, China: Big Data Analytics for Cyber-Physical Systems in Smart
Cities, 28–29 December 2019. Springer 2020, Singapore

Open Access This chapter is licensed under the terms of the Creative Commons Attribution-
NonCommercial 4.0 International License ([Link]
which permits any noncommercial use, sharing, adaptation, distribution and reproduction in any
medium or format, as long as you give appropriate credit to the original author(s) and the source,
provide a link to the Creative Commons license and indicate if changes were made.
The images or other third party material in this chapter are included in the chapter’s Creative
Commons license, unless indicated otherwise in a credit line to the material. If material is not
included in the chapter’s Creative Commons license and your intended use is not permitted by
statutory regulation or exceeds the permitted use, you will need to obtain permission directly from
the copyright holder.