Stockmarket Analysis Using Map Reduce and Py Spark

See discussions, stats, and author profiles for this publication at: https://www.researchgate.
net/publication/356776423
STOCK MARKET ANALYSIS
Article · July 2021
CITATIONS READS
9 789
6 authors, including:
Marimuthu Muthuvel Palaniswamy Velvadivu

Coimbatore Institute of Technology Coimbatore Institute of Technology
27 PUBLICATIONS 102 CITATIONS 10 PUBLICATIONS 13 CITATIONS
SEE PROFILE SEE PROFILE
Some of the authors of this publication are also working on these related projects:
Data Science View project
Internet of Things View project
All content following this page was uploaded by Marimuthu Muthuvel on 04 December 2021.
The user has requested enhancement of the downloaded file.

e-ISSN: 2582-5208
International Research Journal of Modernization in Engineering Technology and Science
( Peer-Reviewed, Open Access, Fully Refereed International Journal )
Volume:03/Issue:07/July-2021 Impact Factor- 5.354 www.irjmets.com
STOCK MARKET ANALYSIS USING MAPREDUCE AND PYSPARK

P.Kavya*1, S.Saagarika*2, R.T.Subavarsshini*3, C.G.Nivetheni*4
Dr. M. Marimuthu*5, Dr. P. Velvadivu*6
*1,2,3,43rd Year, Msc Data Science (Integrated),Coimbatore Institute Of Technology, Coimbatore, India.
*5,6Assistant Professor, Department Of Data Science, Coimbatore Institute Of Technology,
Coimbatore, India.
ABSTRACT
Forecasting the stock market has become very important in planning business activities. The prediction of stock
price has driven many researches in a variety of disciplines, including computer science, statistics, economics,
finance, and operations research. Recent studies have shown that the enormous amount of online information
that is available in the public domain, such as Wikipedia, the social forums, news from media, have a significant
impact on the investor’s opinion towards the financial markets. The reliability of the computational models on
prediction of the stock market is very important, because it is highly responsive to the economy and may result
in financial losses. In this paper, we have made an extensive analysis on various stocks. First, we have
performed Stock Volatility Analysis on 1000 stock dataset of NYSE. The main contributions in this paper
include the development of a dictionary-based sentiment analysis model for the financial sector, and the
evaluation of the model for scaling the effects of news sentiments on stocks for other markets. By using only the
news sentiments, we have achieved a good accuracy of 70.59% in predicting the trends in short-term stock
price movement.
Keywords: Dictionary Comparison, Financial Market , News Articles , Sentiment Analysis , Stock Price
Prediction.
I. INTRODUCTION
Big data has become a great importance for the expansion of various industries and sectors. It is widely used by
business organizations in the formalization of business ideas and intelligence. Furthermore, it has been utilized
by the healthcare sector to discover important patterns and knowledge so as to improve the modern healthcare
systems. Big data also holds significant importance for the technology, information and cloud computing
sector. The financial and banking sectors are utilizing big data to track the financial market activity. Network
analytics and big data analytics are also used to catch illegal trading in the financial markets. Similarly, financial
institutions, companies, traders and big banks are utilizing big data for generating trade analytics which are
utilized in high frequency trading. It also helped in the detection of illegal activities such as financial frauds and
money laundering.
In this paper, we hope to build a system which analyses various stocks by using various Big Data frameworks.
We are going to use Hadoop MapReduce to find out the top 10 stocks with minimum and maximum volatility
and PySpark to predict the closing price of those stocks using Machine Learning models and Sentiment Analysis
using news sentiments.
II. PRELIMINARY KNOWLEDGE
Some basic knowledge of MapReduce, Hive and PySpark are reviewed in this section.
a) HADOOP MAPREDUCE
Hadoop MapReduce is a software framework for writing applications in an easy manner which can process
enormous amounts of data (multi-terabyte data) in-parallel on large clusters (thousands of nodes) of
commodity hardware in a reliable and fault-tolerant manner. The job of MapReduce is to split the inputdata-set
into some independent chunks which are processed by the Map tasks in a completely parallel manner. The
framework then sorts the outputs from the Maps, which are then given as input to the Reduced tasks. Generally
both the input and the output of the jobs are kept in a file-system. The framework takes care of everything from
the scheduling tasks, monitoring them and re-executing the failed tasks. Typically the MapReduce framework
(compute nodes) and the Hadoop Distributed File System (storage nodes) are running on the same set of nodes.
www.irjmets.com @International Research Journal of Modernization in Engineering, Technology and Science
[1785]
e-ISSN: 2582-5208
This configuration allows the framework to effectively schedule tasks on the nodes where the data is already
there, resulting in very high aggregate bandwidth across the cluster. The MapReduce framework consists of a
single master JobTracker and one slave TaskTracker for each cluster-node. This master is responsible for
scheduling the job’s component tasks on the slaves, monitoring them and re-executing the failed tasks. Then the
slaves execute the tasks as directed by the master. Minimally, the applications will specify the input/output
locations and supply the map and reduce functions via implementations of appropriate interfaces and/or
abstract-classes. These, including other job parameters, contain the job configuration. Then the Hadoop job
client submits the job (jar/executable etc.) and configuration to the JobTracker which then distributes the
software/configuration to the slaves, then scheduling tasks and monitoring them, providing status and
diagnostic information to the job-client. Although the entire Hadoop framework is carried out in Java,
MapReduce applications need not be written in Java.
Fig 1: MapReduce Framework

b) PYSPARK
PySpark is nothing but an interface for Apache Spark in the Python language. It not only allows writing Spark
applications using Python APIs, but also provides the PySpark shell which provides interactive analysis of your
data in a distributed environment. PySpark supports most of Apache Spark’s features such as Spark SQL, Mlib
(Machine Learning), DataFrame, Streaming and Spark Core.
Spark is a leading tool in the Hadoop Ecosystem. MapReduce with Hadoop can only be used for batch
processing and cannot work on real-time data. Spark can work stand-alone or over the Hadoop framework to
leverage big data and perform real-time data analytics in a distributed computing environment. It can support
all sorts of complex analysis including Machine Learning, Business Intelligence, Streaming and Batch
processing. Spark is 100 times faster than the Hadoop MapReduce framework for large scale data processing as
it performs in-memory computations thus providing increased speed over MapReduce.
The big-data era has not only forced us to think of fast capable data-storage and processing frameworks but
also platforms for implementing machine learning (ML) algorithms that have applications in many domains.
With a lot of ML tools available, deciding the tool that can perform analysis and implement ML algorithms
efficiently has been a daunting task. Fortunately, Spark provides a flexible platform for implementing a number
of Machine Learning tasks, including classification, regression, optimization, clustering, dimensionality
reduction etc.
III. STOCK MARKET
a) STOCK MARKET INTRODUCTION
The stock market is nothing but a collection of markets and exchanges where activities such as buying, selling,
and issuance of shares of publicly-held companies occur. These financial activities are conducted through

[1786]
e-ISSN: 2582-5208
regularized formal exchanges or over-the-counter (OTC) marketplaces which operate under a defined set of
rules and regulations. There can be any number of stock trading venues in a country or a region which allow
transactions in stocks and other forms of securities. If he/she says that they trade in the stock market, it means
that they buy and sell shares/equities on one (or more) of the stock exchange(s) that are part of the overall
stock market. The main stock exchanges in the U.S. are the New York Stock Exchange (NYSE), Chicago Board
OptionsExchange (CBOE) and the Nasdaq. These leading national exchanges, along with several other
exchanges operating in the country, form the U.S. stock market. Though it is called a stock market or equity
market and is mainly known for trading stocks/equities, other financial securities such as exchange traded
funds (ETF), corporate bonds and derivatives based on stocks, currencies, commodities, and bonds are also
traded in the stock markets.
b) WORKING
In short, the equity markets provide a safe and controlled environment in which market participants can trade
with more confidence in the shares and other eligible financial instruments, with a zero-or low-risk to act in
accordance with the rules made by the regulator of the exchange act, as well as primary markets and secondary
markets.
The primary market of the stock exchange provides companies with the ability to produce and sell their shares
to the public for the first time, at the initial public offering (IPO) process. This activity will help the companies
to raise the necessary capital from investors. In fact, it means that the company will be broken up into a series
of actions (let's say 20 million shares, and the sale of a portion of the shares (say, 5 million shares to the public
at a price (say, $ 10 per share).
In order to facilitate this process, the company will have a market on which the shares are to be sold. This is a
market that is defined by the securities market environment. If all goes according to plan, the company to be
successful is to sell 5 million shares at a price of $ 10 per share, to raise money for the amount of $ 50 million.
Investors will receive shares in the company, they may be expected to hold it for the required period, and in
anticipation of growth in the price of the stock, and any potential return on investment in the form of dividend
payments. The stock exchange acts as an intermediary in the capital of the process and the compensation for
the services of the company and its financial partners.
c) FUNCTION AND PURPOSE
The stock market is one of the best ways to make money for the business, along with the debt markets, which
are generally more imposing but not publicly traded. It allows companies to become publicly traded, and the
raising of additional capital for expansion by selling shares of the company on the open market. The liquidity of
the stock market offers investors a chance to allow their holders to sell securities, fast, and easy. This is an
attractive feature of investing in stocks, compared to other less liquid investments such as real estate and other
real estate.
History has shown that the value of stocks and other assets is an important part of the dynamics of economic
activity, and can influence or be an indicator of public sentiment. In an economy in which the market has been
booming, is considered to be an emerging market. The stock market is often a leading indicator of a country's
economic strength and development.
d) TAXATION
Taxation is a list of all of the strategies and the profits of the shares, including dividends, and are subject to
different tax rates, depending on the type of the safety, security, and longevity. For the most part, the proceeds
of the investment in the shares will be subject to capital gains tax. In many countries, corporations pay a tax to
the government, and the shareholders pay income tax when they make a profit from owning the shares, which
is also known as "double taxation."
e) BEHAVIOR
Changes in the stock prices are mainly caused by external factors, such as socio-economic conditions, inflation
rates, and exchange rates. Intellectual capital has no effect on the current gain of the shares of the company.
Intellectual capital contributes to the growth of the stock return.

[1787]
e-ISSN: 2582-5208
The Efficient Market Hypothesis (EMH) is a hypothesis in economics, and finance, which states that asset prices
reflect all the information that is currently available. The strong efficient market hypothesis does not explain
events such as the collapse in 1987, when the Dow Jones Industrial Average fell to 22.6 percent —this is the
biggest one-day fall in the United States of America.
This event demonstrated that share prices can go down, even if there is no one clear reason for this. An in-
depth search did not match any 'reasonable' development of events that could explain the attack. (Note that
such events are predicted to be strictly random, although very rarely.) In general, it is also true that many of the
prices (other than those projected to occur "at random"), are not caused by new information. A study about the
fifty largest days to day fluctuations in the U.S. stock prices in the post-war period seems to confirm this.
f) STOCK VOLATILITY
Volatility is defined as the extent to which the price of a security increases or decreases as a result of the win. It
is the risk that is associated with a change in the price of a security and is measured by calculating the standard
deviation of the annual rate of return over a specified period of time. Simply stated, it's a measure of how
quickly the value of an asset or a market is moving.
The variable is usually measured by the standard deviation, or variance. At any rate, the higher the value, the
more volatile are the prices or the returns. This means that it has a high standard deviation it means that the
prices will be spread out over a wide range of products. Conversely, a low standard deviation indicates that the
prices are tightly bound in a narrow range.
● Volatility = swings (varies in a short period of time).
● It is a measure of relative performance of a stock.
● High volatility stocks = more swings and vice versa also.
● Whether the values plummet or rise, the fluctuations depend on a day-to-day basis.
IV. OBJECTIVE OF THE STUDY

❖ Stock Volatility Analysis Using Hadoop MapReduce.
❖ Stock Sentiment Analysis.
❖ Stock Price Prediction.
V. STOCK VOLATILITY ANALYSIS USING HADOOP MAPREDUCE
a) DATASET

[1788]
e-ISSN: 2582-5208
To calculate the value of the stocks, the volatility we have 1000 sets of the New York Stock Exchange stock
information. 1000 CSV files, a CSV file that corresponds to each range. Each of the CSV consists of 7 columns for
the Date, Opening, Maximum, Minimum, Close, Volume, and Ad Connect.
b) METHODOLOGY AND IMPLEMENTATION
The standard deviation of the stock volatility, it is now widely used by businesses to identify stocks with high
return potential. This is a project that will help you get to the top-10 stocks, with a maximum and minimum
volatility. There are three phases of MapReduce jobs:
● Mapper -1 - The first MapReduce job to read in a CSV file, containing the number of mappers on an
equal number of your files. Each mapper will read in the data one line at a time. Each row will contain 7 of the
data, i.e., the Date, Open, High, Low, Close, Volume, and the V Shut. However, the computation of volatility only
Adj Close and date is required, so first mapper will map set filename (Stock Name) when you turn the key, and a
comma-separated-value), date, and adj close value.
● Reducer -1 - In this way, the mapping process is complete, the reducer will receive the file name as the
key and all values associated with the same key as the iterative values. The Reducer-1, will be, in principle, to
compute the Rate of Return for a stock for each month and produce output with key as stock name and single
value containing the Rate of Return of a month.
It is used to set the value of the start date and the end date, it is composed of the loading of the initial values of
the iterable, and the installation starts from the beginning of the month. Then run a While Loop over the others
and take care of the past and current values of each and every moment of each and every month. In fact, the last
value. Return will be calculated each time the month changes. The reducer will produce one key value pair for
every month of a stock.
● Mapper -2 - Next up is the Mapper 2. No significant changes in the data. It will be the core values of the
final output of the Reducer -1, and the same data will be sent to Reducer -2.
● Reducer -2 - It will take the key as the stock name and then iterable values will have the Rate of Return
for all the months for a stock. It will calculate the Volatility of the stock according to the formula, Volatility =
Math.sqrt(calc/(noOfMonths - 1) (in the code) and produce the output with key as stock name and values as
volatility of the stock.
● Mapper -3 - It can be used to sort the data in the figures. It accepts the input data with the same key –
value pair as a reducer-2), and the changes made to the key, a value, and the value for a key in order to sort the
data by the variable, and pass it according to volatility.
● Reducer –3 - The generation of the final solution. This will have input with volatility as key and stock
name as value. This will be in sorted order of volatility as the mapper sends data sorted according to the key to
the reducer. So now the reducer will just take the first 10 and last 10 stocks which will give 10 stocks with
lowest volatility and 10 stocks with highest volatility, respectively, as output. Cleanup method is used to print
the top 10 and bottom 10 stocks as it can be done only after completion of the task.
VI. RESULT AND DISCUSSION
Figure 2: These are the top 10 stocks with maximum and minimum volatility.
[1789]
e-ISSN: 2582-5208
VII. MODEL ARCHITECTURE FLOWCHART
Figure 3: Model Architecture Flow Chart

VIII. STOCK SENTIMENT ANALYSIS
In the financial markets, producing a huge amount of data every second." Financial forecasting problems, such
as stock tickers, portfolio, education, and risk management are often complex details of the interactions, which
are difficult to interpret, or to suggest that, on a purely economic model. The application of deep learning
models for this problem can provide more reliable and useful results than the traditional methods. In
particular, deep learning is to explore and exploit the patterns in the data that is visible in the present financial
and economic theory. Deep learning has a number of advantages in the prediction and classification of
permutations that can be avoided, you can take into account the non-linearities and complex data models and
the input data, and can be expanded to include all possibly relevant features. In [HPW16], Heaton, et al.
presented a variety of deep learning methods with applications in finance, including the show, the model of
selection, autoencoders, and lstm center and warehouse. One of the deep learning applications, that is, to see
what the use of an autoencoder is to replicate, the replication of a market with a subset of the stores. The
consistent nature of these models is a good fit for RNN models such as LSTMs and GRUs. We strive to provide
an extension to the existing research in the field of deep time-series analysis as a financial forecast business
strategy.
a) DATA COLLECTION
We consider two aspects of the ability to predict the movements of the stock market. The first of these is the
social data, reports, and is the second-largest source of information on the historical value of the shares. The
News will be downloaded to the New York Times, as well as historical data has been taken from yahoo finance
for the stock. In the second place, the daily stock exchange, is information that can be collected on a variety of
websites, such as finance.yahoo.com and so on.
b) NEWS COLLECTION
We have collected data from Apple computer, Inc. for the past three years, with effect from January 1, 2013, to
April 2, 2016. This data includes the most important events, items, companies, as well as daily stock prices for
AAPL in the same time period. Daily stock prices are inclusive of six parameters: the Open, High, Low, Close,
Adjusted Close, and Volume. To ensure the integrity of the project as a whole, we have taken into account the

[1790]
e-ISSN: 2582-5208
Individual's closing prices for every day of the stock quote. We've collected data from the largest news
aggregators such as news.google.com, reauters.com finance.уаһоо.com.
c) PRE-PROCESSING
Text data is unstructured data. Therefore, we cannot guarantee that the raw test data for the classifier is the
input data. First, you need to tokenize the document, in words, in order to work at the word-level. Text data
that contains noise words, are not contributing to the classification. In addition, the News can contain data,
numbers, and more space, plans, punctuation, stop words, etc., etc.
We have to clear the data by removing all of those words. In order to do this, we have created our own list of
stop words, which includes specific stop words related to the world of finance, as well as a common English
word. This stop words list contains common words, such as proper names, dates, and geographic currency of
the songs. In addition, to ignore words which are only one or two of the documents, we will look at the
minimum possible number of documents that have to take into account the words that appear in at least three
of the documents. The result is the reduction of a word to redundancy. The stemming process, all the words
that can be replaced by the original version of the word. For example, with the words ‘developed’,
‘development’, ‘developing’ has been reduced to the origin of the word ‘develop’. Some of the preprocessing is
done before applying a polarity detection algorithm and some of them are used in the application of the polarity
detection algorithms.
d) DEEP LEARNING MODEL
In this study, we use an RNN approach, consisting of lstm (deep learning) to center blocks, to explore the
predictive validity of the historical messages of the future of information perception. According to the RNN
method, we consider a document as a series of words. If we are to notate in the same order as in the beginning
of the text, then this word is denoted by the vector w-t, as described in Section 3.1Here, t runs from 0 to a max
sequence length that we define. We introduce a vector of a hidden state as ht , and define a recursion formula as
follows:
ht = f (Whht−1 + Wxxt)
where the function f (x) is the activation function (usually sigmoid function or tanh function), and Wh and Wx
are weight matrices. Repeat with the built-in equalizer. Eq.(2) the maximum sequence length, and, finally, the
application of a soft-max function, which is to be assigned any value between 0 and 1, giving us the probability
of a positive and a negative emotion. The repetition of this process, we update the weight matrices, Wh and Wx.
However, the RNN has the problem of vanishing or diverging in the process. The lstm center to solve this
problem, the introduction of cells, which decide whether or not to remove or store information. There are four
main types of gated cells, that is, an Input gate, a neuron with a self-recurrent connection, a forgot gate and an
output gate.
As explained in Section 3.3, we consider the news to be positive, if ri (t) > 0, and negative if ri (t) < 0.) For the
purposes of this definition, a stock related to the message length is greater than or equal to 50 is included in the
DJ29 in 2013 we get a 16,856 for enjoyment and 17,213 all the negative news.
e) MODEL EVALUATION
The values of the hyperparameters are set as follows: the batch size is 24, the number of LSTM units is 64, the
number of output classes is 2 (positive and negative), the max sequence length is 550 (average length of news),
and the number of training iterations is 400 k. Here, we adopted numbers, which are frequently used in the
study of movie reviews, for the batch size and the number of LSTM units.
Keeping the threshold as 0.5, the positive and negative sentiments of news are calculated and for AAPL news,
we got more sentiments which means APPL stocks might have price hikes.
We show learning curves (loss and accuracy) for our LSTM binary classifier on both the training and validation
sets below:

[1791]
e-ISSN: 2582-5208
This shows that the LSTM is very prone to overfitting similar to the convolutional network - as epoch increases,
so does training accuracy, but validation accuracy stays constant, or validation loss increases.
IX. STOCK MARKET PREDICTION

One of the ways to predict stock is to use the concept of using machine learning algorithms. We use it with the
help of Spark which is one of the major tools of the Hadoop Ecosystem. Spark is an open source, library
program used for big data projects.

[1792]
e-ISSN: 2582-5208
a) DATA COLLECTION
One of the ways to predict stock is to use the concept of using machine learning algorithms. We use it with the
help of Spark which is one of the major tools of Hadoop. Ecosystem. Spark is an open source, library program
used for big data projects. Data collection is one of the most important steps to accurately predict stock market
behavior. There are two ways to collect one data by analyzing daily stock market data and the next is to find
calendar information. Everyday details in the stock market can be collected on many websites like
finance.yahoo.com etc. Also, stock market information can be downloaded through many APIs.
b) PRE- PROCESSING NUMERIC DATA
We have taken the stock prediction dataset which has an independent variable x i.e., Date, Open, High, Low,
Close, Volume and one dependent variable y i.e., LABEL.
c) FEATURE SELECTION
The ISNULL () operator is used to process the return value if the value in a particular row or column is null.
Performs functions such as mean, median, and mode and returns the solution of the sentence if the expression
is not null. But in terms of our data, there is no need to replace null values with new values because the
percentage of lost value is about 2%. We therefore plan to drop the data. Since dumping this 2% data does not
make a change in the model. Therefore 2% of the data is reduced using dropna () function.
Our next step is to choose the attributes that are necessary and then leave the attributes that we don’t need. We
have selected variables based on the output of the correlation matrix. The scale is best used for variables that
show a linear relationship between each other. It can be indicated by a scatter plot. With the help of this, we
have found the required variables. The selected variables are "LABEL", "Open", "High", "Low", "Close",
"Volume", "Interest rate", "ExchangeRate", "VIX", "Gold", "Oil", "TEDSpread". We selected it by using data.loc []
function.
The next process is to split the training and testing data. We are implementing this by using the randomSplit()
function . In this we are splitting the data into the ratio of 7:3.
To improve the preprocessing process, we divide the attributes into 2 sets such as numCols containing
numerical values, and catCols which have categorical variables. We use One hot encoding process in catCols,
which helps us better represent category information. This is because most machine learning algorithms do not
work better with categorical variables, so it is better to encode. One hot code entry will be 0’s and 1’s.It will
make them a standard state to identify whenever it is required or needed while performing.
Label encoding is used to convert categorical values into numbers so that the machine is able to read the data. It
is considered an important step in the preprocessing of structured data. Each machine learning algorithm
determines how well the labels work. In labeling, we will place the variables in categories between 0 values and
(number of classes - 1). Of the selected values, all of them were converted into numerical values except height
which has already consisted of numerical values by using the function LabelEncoder ().
String Indexer helps to provide label indices to machine learning columns. It is nothing but when we give the
input column in numbers it changes that into strings and then it indexes the string values.
To feed the columns into the machine learning algorithms the entire column list needs to be converted to a
single vector value. It is considered helpful in combining various features into a vector of a single element. This
is mainly done to train ML models such as SVM, random forest, Decision Tree etc. It is done using the
[1793]
e-ISSN: 2582-5208
VectorAssembler () function. We are creating a new column named VectorAssembler_feature and inserting the
output into this column.
All the input features provided are converted into a single object using a function called Dense Vector function.
The data framework has 11 columns and features that are now ready are given as input into the machine
learning algorithm.
h) MACHINE LEARNING MODEL:
Machine learning algorithms are performed by Spark which has a function named Mlib. Then we fitted the
Decision tree regressor to the training data set. Here the aim is to find the label values to determine whether
the stock is increasing or decreasing at the end of a particular day. We have chosen this because the results are
better on this when compared with other models and it also reduces the overfitting of training variables. We
have got 90% accuracy by implementing the Decision tree algorithm.
g) MODEL EVALUATION:
In this case, MLs and big data tools have been used for better stock market analysis. Because the stock market is
often done. For evaluating the model, we used 2 metrics, first is a Mean absolute (MAE) error. It is a measure of
the difference between paired observations which is expressed in the same way. It helps to measure the
accuracy of the continuous variance. Our predicted MAE score is 1.024, which is a good score.
The second metric counts the r-square ratio, the ratio that shows the value of the x variance (dependent
variable) which is defined by the independent variable. A value of more than 60% r square is considered
positive. In our model, the score rate is 70%.
X. CORRELATION ANALYSIS OF PRICE AND SENTIMENTS
After training and testing the model, we get coefficients which will tell us basically how the stock market price
is related to the headlines and the historic price of the particular stock. The positive coefficients determine that
the two factors are directly related and negative coefficients tell us that data is inversely proportional. This
correlation model is going to help us to predict the stock price of the new data set.
XI. LIMITATIONS OF THE STUDY
● One of the major limitations of this approach is the changing stock market prices. Although the model
predicts the stock of the next prices, there are additional factors from nature that affect stock prices.
● Using technical analysis to select stocks and increasing your investing potential to invest in which stock
through algorithms, may lead to risk management and might tend to lose your money.
XII. FUTURE SCOPE OF STUDY
● Potential improvements can be made to the data collection and analysis method.
● Future research can be done on potential improvements such as, using more refined data, time frames and
more accurate algorithms that are associated with the new dataset.
● Real time trading model, with live streaming data that can be upgraded to calculate the total returns or
investments in real time. This helps in increasing the efficiency of the model and would boost the accuracy
of the model.
XIII. CONCLUSION
● In Stock Volatility Analysis, being below the bottom of the numerical range, zero is obviously a very special
form of volatility. Volatility is zero if there are no changes in the price (price remains constant). So, the
above mentioned are top 10 minimum stocks whose price would never change, since it is closer to zero.
● Higher volatility means that a collateral value can potentially be spread out over a larger range of values. So,
the top 10 are those maximum stocks whose price of the security can change significantly over a short time
period in either way. A lower volatility means that a collateral value does not fluctuate dramatically, and
tends to be steadier.
● In this case, ML and big data tools have been used for better stock market analysis. Because the stock market
is often uncertain. With this, we are able to avoid the investors from facing significant financial losses.

[1794]
e-ISSN: 2582-5208
● From this, it is clear that this process of forecasting the stock market has been quite difficult and many
preprocess steps have been done to bring high accuracy to the model.
● From the Machine learning predictions, we are able to predict the stock market movements which helps the
investors to invest money at the correct place and in a timely manner. The calculated results show that the
Decision tree works better than the SVM and logistic regression. Apart from these regression methods,
many methods can be implemented in finding better results. This can be achieved by using the Neural
networks, by increasing the number of nodes for better accuracy. And we also found that the Feed-Forward
neural network gives a more accurate prediction of the opening price of the stock.
● This study is also a niche application of sentiment analysis in gauging the effects of news sentiments on
stocks for the pharmaceutical market sector. One major contribution of this work is a sentiment analysis
dictionary.
● The sentiment scores obtained from the analysis of the news articles is a powerful indicator of stock
movements and can be used to effectively leverage the prediction of short-term trends. We believe that the
reason the model is able to achieve an accuracy of 70.59% with the dictionary-based approach is that the
dictionary was specifically created for this particular sector by researching and leveraging domain
expertise.
● By analyzing the results of sentimental analysis and prediction, the stock of Apple is positive and good. We
therefore recommend investors to invest in Apple and make a good profit.
XIV. REFERENCES
[1] SENTENCE EMBEDDING PREDICTION.IPYNB - USED THE EVENT EMBEDDINGS INTO AN LSTM.
[2] HTTPS://GITHUB.COM/SRIZZLE/ DEEP-TIME-SERIES/PAPERS - RELEVANT PAPERS (USED FOR REFERENCES)
[3] HTTPS://EN.WIKIPEDIA.ORG/WIKI/STOCK_MARKET
[4] HTTPS://WWW.INVESTOPEDIA.COM/TERMS/S/STOCKMARKET.ASP
[5] HTTPS://WWW.ANALYTICSVIDHYA.COM/BLOG/2021/01/IMAGE-CLASSIFICATION-USING-CONVOLUTIONAL-
NEURAL-NETWORKS-A-STEP-BY-STEP-GUIDE/
[6] HTTPS://MEDIUM.COM/@RAGHAVPRABHU/UNDERSTANDING-OF-CONVOLUTIONAL-NEURAL-NETWORK-CNN-DEEP-
LEARNING-99760835F148
[7] HTTP://WWW.IRAJ.IN/JOURNAL/JOURNAL_FILE/JOURNAL_PDF/14-481-153485282984-86.PDF
[8] HTTPS://ARXIV.ORG/FTP/ARXIV/PAPERS/1607/1607.01958.PDF
[9] HTTPS://GITHUB.COM/ADITYAJAIN10/STOCK-PREDICTOR-PYSPARK-MLIB
[10] HTTPS://WWW.ANALYTICSVIDHYA.COM/BLOG/2019/12/STREAMING-DATA-PYSPARK-MACHINE-LEARNING-
MODEL/
[11] MULTISTEP PRED.IPYNB - USED VADER SENTIMENT AND PREVIOUS PRICES FOR AN LSTM

[1795]
View publication stats

Stockmarket Analysis Using Map Reduce and Py Spark

Uploaded by

Copyright:

Available Formats

Stockmarket Analysis Using Map Reduce and Py Spark

Uploaded by

Document Information

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Stockmarket Analysis Using Map Reduce and Py Spark

Uploaded by

Copyright:

Available Formats

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

STOCK MARKET ANALYSIS

Article · July 2021

Marimuthu Muthuvel Palaniswamy Velvadivu

SEE PROFILE SEE PROFILE

Data Science View project

Internet of Things View project

The user has requested enhancement of the downloaded file.

STOCK MARKET ANALYSIS USING MAPREDUCE AND PYSPARK

Fig 1: MapReduce Framework

www.irjmets.com @International Research Journal of Modernization in Engineering, Technology and Science

www.irjmets.com @International Research Journal of Modernization in Engineering, Technology and Science

IV. OBJECTIVE OF THE STUDY

www.irjmets.com @International Research Journal of Modernization in Engineering, Technology and Science

Figure 3: Model Architecture Flow Chart

www.irjmets.com @International Research Journal of Modernization in Engineering, Technology and Science

www.irjmets.com @International Research Journal of Modernization in Engineering, Technology and Science

IX. STOCK MARKET PREDICTION

www.irjmets.com @International Research Journal of Modernization in Engineering, Technology and Science

www.irjmets.com @International Research Journal of Modernization in Engineering, Technology and Science

www.irjmets.com @International Research Journal of Modernization in Engineering, Technology and Science

You might also like