See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/352197415

REVIEW OF STOCK PREDICTION USING MACHINE LEARNING TECHNIQUES

Conference Paper · June 2021

DOI: 10.1109/ICOEI51242.2021.9453099

CITATIONS READS

22 3,777

4 authors:

Ramkrishna Patel Vikas Choudhary

National Institute of Technology, Kurukshetra National Institute of Technology, Kurukshetra
3 PUBLICATIONS 39 CITATIONS 3 PUBLICATIONS 39 CITATIONS

SEE PROFILE SEE PROFILE

Deepika Saxena Ashutosh Kumar Singh

The University of Aizu Indian Institute of Information Technology Bhopal
91 PUBLICATIONS 1,438 CITATIONS 481 PUBLICATIONS 6,394 CITATIONS

SEE PROFILE SEE PROFILE

All content following this page was uploaded by Deepika Saxena on 15 July 2021.

The user has requested enhancement of the downloaded file.

 Proceedings of the Fifth International Conference on Trends in Electronics and Informatics (ICOEI).
IEEE Xplore Part Number:CFP21J32-ART; ISBN:978-1-6654-1571-2

REVIEW OF STOCK PREDICTION USING

MACHINE LEARNING TECHNIQUES
RAMKRISHNA PATEL VIKAS CHOUDHARY DEEPIKA SAXENA ASHUTOSH KUMAR
SINGH
Department of Computer Department of Computer Department of Computer Department of Computer
Application Application Application Application
NIT KURUKSHETRA NIT KURUKSHETRA NIT KURUKSHETRA NIT KURUKSHETRA
Haryana, India Haryana, India Haryana, India Haryana, India
[email protected] [email protected] [email protected] [email protected]

Abstract—Stock prices change everyday by market forces lower risk. There are two categories for stock prediction
(supply and demand). In recent years stock price prediction technical and fundamental analysis. Technical analysis deals
has been one of the most significant concern. Investors are with the analysis of historical stock prices to forecast the
investing on stock market on the basis of certain prediction. stock prices. On the other hand, fundamental analysis
For prediction, stock market prices investors are applying usually deals on the unstructured dataset such as financial
some techniques and methods through which they get more
news and economic data. To achieve a better result or
profits and minimize their risks. Machine Learning methods
are often used for the prediction of stock prices. This survey accuracy, most researchers concluded that both structured
paper discusses various machine learning approaches and unstructured data sets are used. Many machine learning
(Supervised or Unsupervised) and methods through which the techniques are used in stock price prediction like Naive
investors get to know the stock prices increase or decrease. It Bayesian, Support Vector Machine, K-Nearest Neighbor,
was done in five phases, such as data acquired, pre-processing Artificial neural network, Random forest and the important
of dataset, extraction of features, prediction of stock price branch of machine learning algorithm, Neural network(NN),
using different techniques and display the result. In first phase, Deep neural networking(DNN). The main aim of this
the data is collected from different social sites, historical data research is to provide a detailed analysis using different
of companies. In second phase, the removal of incorrect,
methodologies of stock market forecasting to predict future
duplicate and dirt is done in pre-processing phase. In third
phase, the reduction of data sets and the selection of useful trends and stock returns. This paper explains many of the
data is done. In fourth phase, prediction is done using different current stock market prediction tools and techniques adapted
machine learning techniques and approaches which is to the recant research works [3]. An overview of how the
categorized as supervised and unsupervised learning stock price predicted from sentiments can be seen in Fig.1.
techniques. Now, in last phase the accuracy is determined using The next section of this paper contains the literature
different approaches. survey of the related work in this field. Section 3 provides
us different exciting stock prediction approaches. Section 4
Keywords—Stock prediction, Machine learning, sentimental summery of existing work for stock prediction carries on
analysis. with the comparative analysis of algorithms along with their
accuracy, followed by Section 5 which contains the
I. INTRODUCTION challenges. In the end, the paper is concluded along with the
The stock market is one of the most important part of every future advances needed in Section 6.
country, It fluctuates the economic growth of the country
and also provide a huge effect on many fields, such as News Feed Historical Stock Price
employment and industry etc. According to research only
10% of people of every country take a risk to invest in the
stock market, it is very difficult to invest because of its Pre-Processing
dynamic nature, non- stationary, noisy and non-parametric
nature [1]. Stock price change due to political problems, the
financial economic crisis and many other factors affecting Feature Extraction
the market. Market forces adjust stock prices on a daily
basis. Stock prices fluctuate as a result of supply and
demand. The price of a stock increases when there are more
Classification
people who want to buy it (demand) than there are people
who want to sell it (supply). If more people wanted to sell a
stock than buy it, the supply would exceed the demand,
Stock Prediction
causing the price to drop [2]. Many researchers are
consistently working on building an efficient, inexpensive,
and feasible model which guides investors, newcomers, and
shareholders in a better way that where to hold or invest Fig.1 Operational flow of Stock Prediction.
their stock so that it can maximize the profitability with

978-1-6654-1571-2/21/$31.00 ©2021 IEEE 840

 Proceedings of the Fifth International Conference on Trends in Electronics and Informatics (ICOEI).
IEEE Xplore Part Number:CFP21J32-ART; ISBN:978-1-6654-1571-2

II. LITERATURE SURVEY market state. Researchers make use of python scripting
Different researchers have used intelligent approaches and language which has a quick implementation context and
strategies in the stock market for decision making in recant this will assist the people(investors) what share capital
year. should be invested in.

Bharathi et al. [3] have proposed a method for stock Ayman E. Khedr et al. [8] stated the stock market is seen as
prediction in which a connection between stock market a treasure of investment information, In this model, The
values and RSS news feed is developed by applying authors uses two types of news content: market relevant
sentiment analysis algorithm such as Natural Language news, business news and financial report written by
Processing (POS taggers , SSS algorithm),lexicon approach( financial analysts on stocks Author first stage analysis, news
i.e. Dictionary Based Approach). The authors have used the statement to get the text polarity by using Naive bayies and
combination of two datasets Really Simple obtains 72.82% accuracy, In second stage author use both
Syndication[RSS] news feeds and sensex points of ARBK datasets news and historical stock data to forecast the stock
from Amman stock exchange for effective prediction. prices using a K - NN algorithm and obtains 89% accuracy,
Experimental output have yielded valuable results, increase Author says this model achieved the highest accuracy
in precision prediction of 14.43 percent, when Compared compared to other previous models, This model predict the
with the regular ID3, C4.5 algorithm. future stock market trend and obtained accuracy up to
89.80%.
Vijaya Kumar B P et al. [4] stated that inexpensive, efficient
model to predict the stock price using sentiment analysis on Prof. G.S. Mate et al. [9] assumed that stock prediction and
social media data these are the crucial factors for applying financial news are interlinked to each other. In this paper,
the ability of machine learning for stock prediction. Main Two types of data are gathered which are stock indices and
focus on the minimum the difference between predicted news data from the NY Times Archive API. The Natural
value given by model and the everyday stock value. Author Language Toolkit package in python is most commonly
discussed about many techniques which are important for used for sentiment analysis to identify emotions or
predicting stocks techniques are: SVM, LSTM, KNN, actions by natural language processing. If the news is
LINEAR REGRESSION. positive, then we can state that this news effect is good on
the stock, so that more chances of stock prices are high. And
Sunil Kumar Khatri et al. [5] have considered that artificial if the news were bad, it might have an effect on the stock
neural network used in problems involving computer tasks, price to go down. Output of sentiment analysis is fed into
research, similarity finding, and much more. Sentimental the machine learning model to forecast stock prices.
evaluation is analyzed in this paper on data extracted from
Twitter. This research article carried out an analysis of the Saloni Mohan1 et al. [10] have developed forecast models
sentiments gathered from Yahoo and prepared artificial based on time series prediction models such as ARIMA,
neural network with performance and market prices for the RNN, and Facebook Prophet. This paper forecasts stock
five largest I.T. Companies estimate the return on prices using time series models and a mixture of neural
investment for the future. networks and news posts. The dataset in this paper contains
Regular prices of stock for S&P500 companies over five
Xiaodong Li a Haoran Xie et al. [6] have introduced a years, along with news articles linked to these businesses.
standardized system for forecasting stock price. By using Authors have achieved better results with RN.
the psychological dictionary of Harvard and Loughran-
McDonald's financial sentiment dictionary to create the Ghaith et al. [11] have focused on the prediction
sentiment dimensions. In this paper, Experiments are classification Model using Twitter Sentiment based on the
performed on five years of historical Hong Kong Stock HNBC(Hybrid Naive Bayes Classifiers). The authors
Exchange prices and news articles. The results show that (1) divided their work into five phases – data collection (Twitter
Models with sentiment analysis outperform the bag-of-word API), filtration and transformation (to obtain the relevant
model at the individual stock, sector and index levels. (2) data), labelling (In which data polarity is specified and
Models using sentiment polarity cannot make useful negative, positive or neutral values are assigned to the
predictions. (3) there is a small difference between models opinion of the individual concerned),
using two separate sentiment dictionaries. classification(acceptable stock market patterns are
established by the hybridization of Bayies Classifiers) and
Aditya Bhardwaja et al. [7] found Sensex and Nifty to be last one is performance and evaluation. The proposed work
two important measures for forecasting the stock market achieved accuracy = 90.38%.
situation in india. By which shareholders and investors are
able to understand the stock market situations that relies on Vaanchitha Kalyanaraman et al. [12] This paper presents a
the routine verification and checking of senex point and sentiment analysis on news articles to see the impact on
Nifty forecast values. This paper analyses stock market stock prices. Dataset was from the Bing API. There is a
sentiment by gathering Sensex and Nifty live server data at specialized sentiment dictionary intended solely for the
various time intervals that can be used to forecast stock. study of stock papers. Two separate machine learning
algorithms were used for the data set and the performance of

978-1-6654-1571-2/21/$31.00 ©2021 IEEE 841

 Proceedings of the Fifth International Conference on Trends in Electronics and Informatics (ICOEI).
IEEE Xplore Part Number:CFP21J32-ART; ISBN:978-1-6654-1571-2

the two was compared. There is a comparison of the Data Pre-processing phase resolves three major type of
projected effects with the real shift in stock prices. issues- Data cleaning, Data Transformation, Data
Reduction. In feature selection phase of machine learning
Rakhi Batra et al. [13] considered SVM models for stock the reduction of datasets and selection of useful data is done.
prediction because they can handle large feature spaces and There are two major types of feature selection techniques:
are also effective for text categorization. Authors uses two supervised and unsupervised methods. When developing a
dataset tweets related to Apple products and market index predictive model, feature selection is the method of reducing
data from Yahoo Finance. The results of the proposed work the number of input variables. The number of input
are 75.22 percent accuracy of the training and 76.68 percent variables should be reduced to reduce the computational
accuracy of the test. According to the authors, this work cost of modeling and, in some cases, to increase the model's
needs to improve the accuracy of the text classification test accuracy.
and the accuracy of the prediction if the size of the dataset is
increased. Natural Language Processing (NLP) [10]: opinion mining
is a methodology use to determine the mood, opinion and
Dev Shah et al. [2] focused on some of the stock market emotion of an individual about and specific issues . To
forecasting research achievements and discussed some of recognize and bring out subjective information it utilize the
the forecasting methods and technical, fundamental, short- processing of natural language and text analysis in source
term, long-term approaches that help predict stock prices. material.
According to the author, a combination of statistical and
machine learning techniques is likely to be more useful for Dictionary based Approach [3] [4]: Dictionary-based
stock prediction, The authors note that the market acts as a methods use synonyms, antonyms and hierarchies WordNet
weighting machine with less noise and more predictability to decide the meaning of the word.
in the long run.
Naïve Bayes [3]: Naive Bayes has been one of the easiest
Antony et al. [14] considered that ANN (Artificial Neural and most reliable classification machine learning
Network) was a highly versatile, non-linear approach to technique that strengthens fast machine learning models that
forecasting stock prices for BSE-related companies can make quick predictions. It is a probabilistic classifier,
(Bombay Stock Exchange). In this model, the author uses which implies that it predicts on the basis of the probability
the opening price, high, low, closing price, and stock of an object.
volume as a dataset. According to the author, if we want to
increase the accuracy rather than increase the number of Support Vector Machine (SVM) [17]: Support Vector
inputs, because the author observed that the error in the Machine(also known as support-vector networks) is a set of
model was high when they took three inputs for prediction, related supervisory techniques used to classify and regress.
and it was minimized when the five inputs and the accuracy SVM is a supervised machine learning models with
of the predictions were higher. associated learning algorithms that analyze data for
classification and regression analysis in machine learning.
Ryo Akita et al. [15] proposed Deep Learning Models and SVMs, which are based on mathematical learning models,
LSTM for prediction of stock prices. Authors approach are one of the most reliable prediction methods [18].
predicts stock prices by using Regression for textual
information and LSTM for numerical information. Paper LINEAR REGRESSION [8]: Linear Regression is a part
results have shown that the distributed representation of text of machine learning algorithm(supervised learning). It's
data is better than the numerical data and Bag-of-Words performing a regression operation. It is often used to assess
based methods, LSTM was effective for time series data the relationship between variables and forecasting.
impacts than other models.
K-nearest neighbor (KNN) [7]: The method for
Jing Zhang et al. [16] proposed a new stock market trend information processing is k-nearest neighbor. In machine
prediction method that can forecast financial performance as learning methods, K nearest neighbor is a simple algorithm
well as its interval of increase (or decrease) rate within where the categorization is accomplished by recognizing the
predefined forecasting periods. According to this paper, the closest neighbor to query examples and using those
model uses an heuristic algorithm (unsupervised) to split neighbors to determine the question class.
raw user information from each stock into multiple clips of
predetermined duration and classify them into four Artificial Neural Network (ANN) [19] [20]: is a
categories Up, Down, Flat, and Unknown. This proposed component of a computer system designed to stimulate the
model outperforms some existing techniques. gathering and processing of human brain information. It is
the backbone of Artificial Intelligence (AI) and resolves
issues that would prove impossible or difficult by human or
III. STOCK PREDICTION APPROACHES statistical standards [21]. ANN is commonly use for
Pre-Processing and Feature Selection [3]: In machine regression as well as classfication in various field [22].
learning, pre-processing is a crucial phase to get rid of the
repetitive, outdated, and incorrectly formatted content. Dirty Long short team memory (LSTM) [23]: is an
data may cause uncertainty in the third step. As a result, data RNN(recurrent neural network). LSTM networks used for
pre-processing is carried out in order to clean the dataset. forecasting based

978-1-6654-1571-2/21/$31.00 ©2021 IEEE 842

 Proceedings of the Fifth International Conference on Trends in Electronics and Informatics (ICOEI).
IEEE Xplore Part Number:CFP21J32-ART; ISBN:978-1-6654-1571-2

on real time series data, classification of data processing of data [24].

IV. SUMMARY OF EXISITING WORK FOR STOCK PREDICTION.

Year Literature Future selection Classification Dataset Result (accuracy) Limitation

reference Approach (Approach)

2013 Antony et al ANN(Artificial Supervised BSE(Bombay stock Improve the accuracy Difficulties in
neural network) approach exchange) and transmitting the issue to
news article. the network.

2014 Vaanchitha Sentiment Supervised News articles predicted sentiment

Kalyanarama dictionary learning using linear regression If the sentimantal
n et al. with normal dictionary does not
equation.=54.54% and recognise a word, it
with Gradient must be taken in
Desent=81.81% isolation.

2014 Xiaodong Li psychological Bag-of-words news articles and LMD achieves the best The sense of the word
a Haoran Xie dictionary of approach historical Stock in both in the text is ignored by
et al. Harvard and Exchange prices. validation(0:5976) and the approach. It pays
Loughran- testing(0:5527) when little attention to the
McDonald's compared to six context in which it is
financial different approaches. used.
dictionary.
2015 Aditya Python scripting Sensex data and Prediction of stock Third-party frameworks
Bhardwaja* language. Nifty data values. market prices. and libraries are
et al. needed.

2016 Sunil Kumar NLP and Text Supervised Stock Twits Value determined for Lack of research and
Khatri1 et al. Analysis approach each company with a development.
(artificial neural mean square error
network) Low-resource language

2016 Ghaith Hybrid Navie Supervised Twitter Dataset Accuracy = 90.38% . This limits the
Abdulsattar Bayes Classifiers. approach (Twitter API). applicability of this
A. Jabbar algorithm in real-world
Alkubaisi et use cases.
al. In certain cases, its
predictions are
incorrect.
2017 Shri Really Simple Accuracy improve Since the input data is
Bharathi.Sv Natural Language Unsupervised Syndication[RSS] 14.43% as compared unknown and not
et al. Processing, learning news feeds and to ID3, C4.5. labelled in advance, the
lexicon sensex points of results are less
approach(Dictionar ARBK accurate.
y Based
Approach)

978-1-6654-1571-2/21/$31.00 ©2021 IEEE 843

 Proceedings of the Fifth International Conference on Trends in Electronics and Informatics (ICOEI).
IEEE Xplore Part Number:CFP21J32-ART; ISBN:978-1-6654-1571-2

2018 Rakhi Batra SVM(Support Supervised Market index data Accuracy up to Large datasets require a
et al. vector machine ) Approach. and tweets from 76.65% long training time.
twitter

2018 Jing Zhang et Heuristic Unsupervised Stock Price Data Improves the When approximate
al. algorithm learning accuracy. solutions are sufficient
but exact solutions are
computationally costly,
heuristic algorithms are
frequently used.
2019 Saloni Regression, Time Supervised Daily stock prices Improve the accuracy Only LINEAR
Mohan1 et al. series models, approach (Deep and news articles. of stock price relationships are being
Neural Network Learning). predictions. considered.
and financial
news articles
2019 Dev. Shah et combination of Supervised Improve the accuracy. Needs lot of time.
al. statistical and approach Daily stock price
machine learning and news articles.
techniques

2019 Prof. G. S. natural language Unsupervised Stock indices and Predict the stock Machines have a hard
Mate el al. processing Machine news data from the prices time implementing
learning NY Times Archive NLP.
API

2019 Ayman E. Naive bays market relevant Accuracy up In certain cases, its
Khedr et al K - NN algorithm. Supervised news, business to 89.80%. predictions are
approach news incorrect.

2020 VijayaKumar k-NN algorithm, Supervised News articles and Accuracy up to 85% Probability estimation
et al. regression approach historical data would be zero if there
analysis are no appearances of a
class label.
2021 Ryo Akita et LSTM(long short bag-of-words miscellaneous Predict the stock The sense of the word
al. term memory) approach events reported in prices. in the text is ignored by
newspapers the approach. It pays
little attention to the
context in which it is
used.

V. CHALLENGES:
(3) Selecting the appropriate machine learning model for
(1). Stock Prediction is dynamic in nature. the respective prediction.
This is due to the fluctuations in stock market which Suppose If the training data is much larger than no. of
changes every day. To overcome this challenge we need to feature in such case, KNN is better than SVM but if no. Of
Re-train online our model periodically so that we get to features is much larger than training data, then SVM
know the prediction of every change. outperforms KNN.
(2) Re-training the model frequently to predict the stock (4) Out of vast the real time amount of dataset sorting,
market change. pre-processing, features extraction is difficult task to
There is an ambiguity that after how much period we need evaluate.
to Re-train our model so that we get optimum prediction.

978-1-6654-1571-2/21/$31.00 ©2021 IEEE 844

 Proceedings of the Fifth International Conference on Trends in Electronics and Informatics (ICOEI).
IEEE Xplore Part Number:CFP21J32-ART; ISBN:978-1-6654-1571-2

VI. CONCLUSION AND FUTURE WORK: [9] G. S. Amidwar, R. Kulkarni and M. Muthya,
"Stock Prediction Through News Sentiment
The goal of our research study is to help stockbrokers and Analysis," Journal Of Architecture & Technology,
investors to invest capital in the stock market. Prediction vol. 9, pp. 36-40, 2019.
plays vital role in the stock market, which is a very
complicated and difficult process due to the political
problems, the financial economic crisis and many other [10] S. Mohan, S. Mullapudi, S. Sammeta, . V. Parag
factors affecting the market. This survey paper discusses and . D. C. Anastasiu, "Stock Price Prediction
various machine learning approaches such as NLP, Linear Using News Sentiment Analysis," International
Regression, KNN, SVM, LSTM, Artificial Neural Conference on Big Data Computing Service and
Networking and many more. Merit of prediction model is to Applications, pp. 205-208, 2019.
guides investors, newcomers, and shareholders in a better
way that where to hold or invest their stocks so that it can [11] G. Abdulsattar, S. S. Kamaruddin and H. Husni,
maximize the profitability with lower risk. On the other "Stock Market Classification Model Using
hand, If the dataset contains fake news and inappropriate Sentiment Analysis on Twitter Based on Hybrid
information then the prediction of stock price would be Naive Bayes Classifiers," Computer and
incorrect. Our plan for future would be to improve the Information Science, vol. 11, pp. 52-64, 2018.
accuracy of stock prediction by implementing the new
method. We will prepare a model which will be more [12] V. Kalyanaraman, S. Kazi , T. Rohan and S.
accurate then the existing one and it will also overcome the Oswal , "Sentiment Analysis on News Articles for
current limitations. Stocks," Asia International Conference Modelling
Symposium, pp. 10-15, 2014.
References
[13 R. Batra and S. M. Daudpota, "Integrating
StockTwits with Sentiment Analysis for better
[1] A. Sharma, D. Bhuriya and U. Singh, "Survey of
Prediction of Stock Price Movement,"
Stock Market Prediction Using," International
International Conference on Computing,
Conference on Electronics, Communication and
Mathematics and Engineering Technologies –
Aerospace Technology, pp. 506-509, 2017.
iCoMET, vol. 2, 2018.
[2] D. Shah, H. Isah and F. Zulkernine, "Stock Market
[14] A. V. Devadoss and T. A. A. Ligori, "Stock
Analysis: A Review and Taxonomy of Preddiction
Prediction Using Artificial Neural Networks,"
Techniques," International Journal of Financial
International Journal of Data Mining Techniques
Studies, pp. 2-22, 2019.
and Applications, vol. 02, pp. 283-291, 2013.
[3] S. Bharathi and A. Gheetha, "Sentiment Analyssis
R. Akita, A. Yoshihara, T. Matsubara and K.
for Effective Stock Market Prediction,"
[15] Uehara, "Deep Learning for Stock Prediction
International Journal of Intelligent Engineering
Using Numerical and Textual Information,"
and Systems, pp. 142-153, 2017.
Authorized licensed use limited to: IEEE Xplore,
2016.
[4] N. N. Reddy, V. K. B P and N. E. , "Prediction
Stock Price Using Sentimental Analysis Through
[16] J. Zhang, S. Cui, Y. Xu, Q. Li and T. Li, "A novel
Twitter Data," UTC from IEEE Xplore , 2020.
data-driven stock price trend prediction system,"
Expert Systems With Applications, pp. 60-69,
[5] S. K. Khatri and A. Srivastava, "Using Sentimental 2018.
Analysis in Prediction of Stock Market
Investment," International Conference on
[17] L. and M. Chi, "Using support vector machine
Reliability, Infocom Technologies and
with a hybrid feature selection method to the stock
Optimization, pp. 566-563, 2016.
trend prediction," Expert Systems with
Applications 36 , vol. 8, pp. 10896-10904, 2009.
[6] X. Li and J. Wang, "News impact on stock price
return via sentiment analysis," Knowledge Based
[18] Guo, . H. A. Zhang and W. Wang, "An accelerator
System, vol. 69, pp. 14-23, 2014.
for online SVM based on the fixed-size KKT
window," Engineering Applications of Artificial
[7] A. Bhardwaj, Y. Narayan, P. and V. , "Sentiment Intelligence 92, 2020.
Analysis for Indian Stock Market Prediction Using
Sensex and Nifty," Procedia Computer Science,
[19] D. Saxena and A. K. Singh, "A proactive
vol. 70, pp. 85-91, 2015.
autoscaling and energy-efficient VM allocation
framework using online multi-resource neural
[8] A. E. Khedr and N. Yaseen, "Predicting Stock network for cloud data center," Neurocomputing
Market Behavior using Data Mining Technique and 426, pp. 248-264, 2021.
News Sentiment Analysis," Intelligent Systems and
Applications, vol. 7, pp. 22-30, 2017.
[20] D. Saxena and A. K. Singh, "Auto-adaptive
learning-based workload forecasting in dynamic
cloud environment," International Journal of
 Proceedings of the Fifth International Conference on Trends in Electronics and Informatics (ICOEI).
IEEE Xplore Part Number:CFP21J32-ART; ISBN:978-1-6654-1571-2

Computers and Applications , pp. 1-11, 2020.

[21] D. Saxena, A. K. Singh, A. Mohan and J. Kumar,

"Biphase adaptive learning-based neural network
model for cloud datacenter workload forecasting,"
Soft Computing, pp. 1-18, 2020.

[22] D. Saxena, A. K. Singh and R. Buyya, " An

Online VM Prediction based Multi-objective Load
Balancing Framework for Resource Management
at Cloud Datacenter," IEEE Transactions on
Cloud Computing , 2021.

[23] G. D. and . L. QIN, "Study on the prediction of

stock price based on the associated network model
of LSTM.," International Journal of Machine
Learning and Cybernetics 11, vol. 6, pp. 1307-
1317, 2020.

[24] J. Y. Yang and . Y. Liu, "Stock closing price

prediction based on sentiment analysis and
LSTM," Neural Computing and Applications, pp.
1-17, 2019.

[25] T. S. Kumar , "Data Mining Based Marketing

Decision Support System Using, Hybrid Machine
Learning Algorithm" Journal of Artificial
Intelligence 2, vol. 3, pp. 185-193, 2020.

[26] Vivekanadam B. , "Analysis of Recent Trend and

Applications in Block Chain Technology ," Journal
of ISMAC 2, vol. 4, pp. 200-206, 2020.

978-1-6654-1571-2/21/$31.00 ©2021 IEEE 846

View publication stats