Ieee A
Ieee A
Ieee A
Abstract – The key objective of a successful stock market sentiment of news stories, providing a valuable tool amidst the
prediction strategy is to not only generate the highest possible returns growing popularity of these strategies Consequently, it has become
but also to minimize inaccuracies in stock price estimations. In easier to comprehend the evolving trends in the stock market,
trading, utilizing sentiment analysis helps investors make well- offering potentially profitable returns with minimal effort [3].
informed choices about where to put their money. However,
forecasting stock prices is a complex task due to their susceptibility to The realm of stock market analysis stands as a dynamic and
a wide array of influences, including shifts in investor mood, economic pivotal area of inquiry, where the quest for forecasting its
and political landscapes, leadership transitions, and more. Predictions behaviours is critically vital in the contemporary era. The stock
based solely on past data or textual content tend to be unreliable. To market is a very dynamic and uncertain field, so the stock market's
improve accuracy, there's a growing focus on integrating the prediction naturally becomes a burning topic. The inherent
sentiment from news sources with existing stock price information. A complexity of predicting market fluctuations necessitates a deep
deep learning method has been developed to track the trends of and comprehensive examination of data patterns. To tackle this
Nifty50 stocks, utilizing data scraped from social media platforms like complexity, a blend of specialized statistical methodologies and the
Twitter, Facebook, StockTwits, and YouTube. This data was cleaned prowess of artificial intelligence becomes indispensable, guiding us
and analyzed to obtain subjectivity and polarity scores, reflecting toward more precise outcomes. The employment of a spectrum of
positive, neutral, or negative sentiments. By integrating these
machine learning and deep learning techniques holds the promise
sentiment scores with market data, a novel approach was formed to
predict Nifty50 returns using the deep learning model.
of delivering robust predictions characterized by reduced margins
Keywords— Socia media, sentiment analysis, deep learning, stock of error. These advanced computational approaches, by analyzing
movement prediction historical data and identifying underlying patterns, enable
stakeholders to make more informed decisions with a higher degree
I. INTRODUCTION of confidence [4]. The convergence of AI and ML in forecasting
stock market trends marks a significant shift towards a more
A. Background analytical and data-driven approach to stock trading, diminishing
As communications technologies have advanced and high- the reliance on speculative guesswork. As computational
speed internet has become more accessible globally, a diverse array technology runs forward, it promises to unlock even more
of individuals from various backgrounds and cultures has sophisticated AI and ML capabilities, potentially elevating the
increasingly engaged with social media. The ubiquitous presence of precision and efficiency of stock market predictions to
the internet has made social media Networks, blogs, Facebook, and unprecedented levels [5] This technological advancement is not
Twitter very popular and effective. People interact and share their only advantageous for individual traders and financial institutions
ideas, opinions, interests, and personal information [1]. These but also plays a critical role in enhancing the stability and
social media channels have profoundly altered how people transparency of financial markets on a global scale.
communicate and collaborate. However, manually analyzing the The abundance of data sources enhances the depth of
vast volume of user-generated data has become cost-prohibitive, understanding in stock market analysis, leading to more accurate
leading to the development of automated systems like Sentiment stock price predictions than possible. Certain techniques establish
Analysis [2]. Sentiment Analysis can swiftly determine the overall connections between historical data and future stock price
movements, utilizing past trends to forecast upcoming changes [6]. possible for online textual content to mirror investor sentiment and
Stock investors use market trend forecasts to decide the best times forecast trends in the stock market. Consequently, there is a need
to buy or sell stocks, aiming to buy low and sell high to maximize for an efficient approach to derive insights from the vast volume of
profits [7] . However, accurately predicting the stock market is textual documents available [16]. Techniques like sentiment
challenging due to many factors, including the effects of social analysis, opinion mining, natural language processing, and data
media. These variables can significantly impact market trends mining are employed to extract perspectives, feelings, and opinions
positively or negatively, making them crucial for investors to from text-based content [17]. A vital component of these prediction
consider for successful market predictions (Khan et al., 2022). methods is Sentiment analysis, often referred to as opinion mining,
that is a process used in natural language processing and text
Navigating the stock market's volatility requires a disciplined
analysis to systematically identify, extract, quantify, and study
approach for investors seeking substantial returns. Before
affective states and subjective information. It is commonly used to
investing, diligent evaluation of a company's market performance
determine the sentiment or emotional tone behind words in a text
is crucial, which often involves analyzing its presence on social
and to understand the attitudes, opinions, and emotions expressed
media and financial news platforms. However, the sheer volume of
[18],[19] specifically applied these sentiment analysis techniques to
data available from these sources exceeds what investors can
analyse a Twitter dataset, an endeavour that presents unique
feasibly process on their own, underscoring the need for automated
challenges due to the informal, concise, and dynamic nature of
decision support systems. Such systems leverage machine learning
language on social media. Their work serves as an extensive guide
algorithms to sift through vast datasets, identifying trends and
to sentiment analysis within the realm of NLP, showcasing how
making predictions about stock performance [9]. The quest to
different methods can be leveraged to interpret sentiments
pinpoint the most effective algorithms for analyzing external data
expressed in the vast and varied Twitter landscape [21] The study
sources, such as financial news and social media, is critical.
by Singh in 2020 emphasizes the importance of method selection
Accurate predictions based on these external factors can
based on project needs, considering factors like computational
significantly enhance investors' profits, sparking considerable
resources, real-time processing capabilities, and desired accuracy
interest among machine learning researchers dedicated to
levels. This approach underlines the nuanced considerations
improving stock market investment strategies. The surge in
required to effectively employ sentiment analysis on social media
popularity of advanced analytical strategies has significantly
data, highlighting the evolving nature of language and sentiment
enhanced the clarity and comprehensibility of stock market trends.
expression online [21] .
These methods offer a commendable return on investment with
minimal effort required from the investor's side. Given the inherent Weng study created an advanced financial system capable of
dynamism of the stock market, where prices and trends are in assessing the sentiment scores of news articles related to specific
constant flux, the ability to accurately forecast future movements of stocks, to predict short-term stock price movements [22] .
stock prices becomes paramount [10] This necessitates not only a
deep understanding of the market's historical and current II. RESEARCH METHODOLOGY
behaviours but also an adeptness at employing sophisticated
prediction tools and models. These advancements have made it A. Data Gathering
increasingly feasible for investors to navigate the complexities of Secondary Data for the proposed study was extracted from
the stock market, thereby democratizing access to strategies that Yahoo Finance and four social media platforms, including Twitter,
yield respectable returns with reduced effort and risk. StockTwits, Facebook, and YouTube, for four years from 2018 to
2021. Numerous feeds are generated daily on these social media
B. Literature Review platforms about Nifty50 by different market experts. Therefore, the
Various researchers have developed methodologies to enhance feeds have been aggregated and aligned day-wise, corresponding to
the accuracy of stock market predictions, employing a range of the market return date for the sentiment score calculation. The daily
approaches. Jayanth Balaji explored the efficacy of 14 different returns based on the closing price of the Nifty50 were taken as the
deep learning models in forecasting the stock prices of companies, dependent variable to observe the trends corresponding to derived
demonstrating the potential of deep learning in financial sentiments and market positions of Nifty50, including open, high,
predictions [11] . Similarly, Tsong Wuu Lin focused on leveraging low, adjusted close, and volume as the independent variables. The
Artificial Neural Networks (ANN) to optimize profitability, data for Twitter (tweets) has been collected via Twitter API.
showcasing the capability of ANN in financial modelling Octoparse and Facepager were used to manage the data for
[12].Autoregressive models are highlighted for their robustness in StockTwits and Facebook, respectively. Video Transcripts were
stock market forecasting, offering valuable insights into time series taken for data collection on YouTube.
analysis and yielding precise predictions. Additionally, sentiment
analysis has emerged as a powerful tool for stock market B. Data Cleaning and Pre-processing
forecasting, with social media analytics playing a crucial Data preprocessing and cleaning are essential steps in preparing
role [13].The ARIMA model, in particular, is noted for its the dataset for the development of machine learning models for
effectiveness in sentiment analysis and in predicting time series trend prediction of Nifty50. The dataset comprised 20,867 samples
data, underscoring the diverse methodologies researchers are with 15 attributes. Collinearity among predictor variables was
implementing to tackle the dynamic challenge of stock market assessed to potentially impact model performance and
prediction [14]. Deng study employed sentiments from individual, interpretability [24]. Using Kendall's and Spearman's rank
institutional, and foreign investors as predictors for the directional correlation coefficients, a correlation matrix was computed,
trends of the Shanghai Stock Exchange index [15]. The rapid revealing high collinearity between attributes ('Open', 'High', 'Low',
advancement of the Internet, particularly social media, has made it
'Adjusted Close'). To address this, a new feature, 'New feature', was relationships, making them versatile for modeling diverse
engineered to represent the average of these highly correlated behaviors [34].
attributes [25]. Outliers were detected and removed using Z-scores,
The SVM algorithm is widely used in machine learning as it
which measure the deviation of each data point from the mean in
can handle both linear and nonlinear classification tasks. However,
terms of standard deviations [26].
when the data is not linearly separable, kernel functions are used to
( X−μ ) transform the data to higher-dimensional space to enable linear
Z= (1)
σ separation. This application of kernel functions can be known as
the “kernel trick”, and the choice of kernel function, such as linear
Where: X is the data point , μ is the mean of the sample & σ is the kernels, polynomial kernels, radial basis function (RBF) kernels, or
standard deviation of the sample sigmoid kernels, depends on data characteristics and the specific
Data points exceeding a threshold of 3 standard deviations were use case .
considered outliers and subsequently removed from the dataset So, to separate the multi-dimensional data we use hyperplane.
resulting in (20,589 x 5) attributes [27]. To define a hyperplane for two-dimensional data which can be
The dataset was then standardized using the StandardScaler to linearly separable by a line.
ensure uniformity in feature scales, preventing any particular Now we are renaming x with x1 y with x2 then we get:
feature from dominating the modelling process [28].
a x 1−x 2 +b=0 (3)
if we define x = (x1 , x2) and ω = (a ,1 )
Finally, the pre-processed dataset was split into training and testing
sets, with 70% allocated for training and 30% for testing,
facilitating model development and evaluation [29].
ω . x +b=0 (4)
C. Data Preparation
After obtaining the hyperplane, we utilize it for making
A total of 20930 samples were collected to build a (20930, 7)
predictions. We define the hypothesis function h as:
matrix with input and output features for the deep-structured
{
classifier. The input characteristics were scaled via z-score
normalization [30], as mathematically represented in Equation 1: h= +1if ω . x+ b≥ 0
−1 if ω . x +b< 0 (5)
I s=( I a −µ ) / χ (2) The point above or on the hyperplane will be classified as class
+1, and the point below the hyperplane will be classified as class -
Where Ia and Is represent the actual and scaled data, the terms,
1.
µ and χ denote the sample mean and standard deviation. The
scaled dataset was divided into 80% training, 10% validation, and b) Logistic Regression
10% test sets. Finally, the scaled data was reshaped as (BS, 6, 1) to
obtain an acceptable input to the LSTM layer. The output labels Logistic regression, pioneered by David Cox, models the
include two categories, negative and positive trends of the Nifty50 relationship between multiple independent variables and a
stock. dependent variable, specifically suited for situations with binary
outcomes and continuous predictors [35].
D. Machine Learning Unlike traditional regression methods, logistic regression is
Machine learning techniques offer a systematic approach to adept at classifying observations into distinct categories, relaxing
analyzing large volumes of historical market data and identifying assumptions like normality of independent variables and absence of
patterns that may influence future trends. By employing ML multicollinearity [36].
algorithms, we aim to enhance our understanding of the underlying
dynamics driving Nifty50 movements and develop predictive The logistic regression equation is represented as:
models capable of forecasting market trends with greater accuracy
[31].
a) Support vector Machine
(
P Y=
1
)
=
X 1+e
1
−w ⋅ x−b (6)
By combining the predictions of multiple trees, Random Forests 1. Tracking according to social hierarchy
can capture complex relationships between input variables and the To mathematically model the social hierarchy of wolves when
target variable, making them suitable for capturing the nonlinear designing GWO, we consider the fittest solution as the alpha (α).
dynamics [42] , [43]. Consequently, the second and third best solutions are named beta
(ẞ) and delta (δ) respectively. The rest of the candidate solutions
Hyperparameters are used in random forests to either enhance
are assumed to be omega (ω). In the GWO algorithm the hunting
the performance and predictive power of models or to make the
(optimization) is guided by a, β, and δ. The w wolves follow these
model faster. The hyperparameters used by random forest classifier
three wolves.
are n_estimators, max_features , mini_sample_leaf, criterion and
max_leaf_nodes for increasing predictive power . To increase the 2. Encircling the prey
speed n_jobs, random_state, oob_scores are used [44].
As mentioned above, grey wolves encircle prey during the hunt.
d) Gradient Boosting Classifier In order to mathematically model encircling behaviour the
following equations are proposed:
Gradient Boosting Classifier (GBC), pioneered by Jerome
Friedman, is an ensemble method for regression and classification. ⃗ ⃗ .⃗
D=¿ C X (t )− ⃗
X ( t )∨¿ (9)
It iteratively improves the model by combining weak learners and
minimizing a loss function through gradient descent. [45]. ⃗
X ( t+ 1 )=⃗
X p ( t )−⃗
A.⃗
D (10)
Where t indicates the current iteration, ⃗
A and ⃗C are coefficient Flowchart
vectors, ⃗
X p is the position vector of the prey, and ⃗
X indicates the
⃗
position vector of a grey wolf. The vectors A and \overline⃗ C are Begin
calculated as follows:
⃗
A=2∗a . r⃗ 1−a (11) Initialize the population
and calculate the fitness
⃗
C =2 . ⃗r2 (12) value
⃗ ⃗2 . ⃗
D β=¿ C X β− ⃗
X ∨¿ (14) i =i+1
⃗ ⃗ ⃗ ⃗
Dδ =¿ C3 . X δ− X ∨¿ (15)
Update parameters
accordingly
⃗
X ₁= ⃗ X α −⃗A ₁. ⃗
Da (16)
⃗ ⃗ ⃗ ⃗
X 2 = X β− A ₂. Dβ Calculate the objective function
value for all current
(17)
⃗
X ₃= ⃗
X δ− ⃗
A ₃. ⃗
Dδ (18) Update position based on
optimal objective function
⃗
X 1 +⃗
X 2 +⃗
X3
⃗
X ( t+ 1 )= (19) i=i+1
3
4. Attacking prey(exploitation)
To model the wolves' approach to the prey, the algorithm F. Model Paradigm
decreases the value of a, representing the fluctuation range of A, The framework proposed in this study, as depicted in Fig. 1,
from 2 to 0 across iterations. ˉAˉ becomes a random value in the outlines a schematic flow diagram for predicting trends in the
interval [-a, a]. When random values of A fall within [-1, 1], a Nifty50 indices. The input and output labels are fed into a
search agent's next position can be anywhere between its current predictive model, which undergoes preprocessing. A notable
position and the prey's position. Figure 5(a) illustrates that when preprocessing step involves addressing multicollinearity, where
<1A<1, wolves are directed to attack towards the prey. rows with a threshold greater than 0.75 are averaged out into a new
column, effectively treating them as a new feature.
5. Searching for prey(exploration)
In the Grey Wolf Optimization (GWO) algorithm, randomness Fig. 1 Flow diagram of the proposed framework.
is introduced through parameters like ∣A∣ and ∣C∣ to encourage
divergence among search agents, promoting global exploration. ∣A ∣
facilitates exploration, while ∣C∣ provides random weights to
influence prey factors, aiding in avoiding local optima. Unlike ∣A ∣,
∣C∣ maintains randomness throughout optimization, preventing
stagnation in local optima. GWO starts with a random population
of wolves, iteratively estimating prey positions by alpha, beta, and
delta wolves while adjusting their distances accordingly.
6
99.858
RFC - GWO 99.7572 99.7174 0.8948
5
91.626
GBC 79.1456 76.6446 0.8948
5
GBC - GWO 100.00 100.00 100.00 1.00
(a)
(b)
Fig. 2 Confusion matrix for total data (a) SVM without GWO, and
(b) SVM with GWO .
(b) (b)
Fig. 4 Confusion matrix for total data (a) RFC without GWO, Fig. 5 Confusion matrix for total data (a) GBC without GWO, and
and (b) RFC with GWO . (b) GBC with GWO .
The confusion matrices shown in fig 4 for Random Forest
represent the classification outcomes both before and after the The evolution of the Gradient Boosting Classifier (GBC) can be
integration of Grey Wolf Optimization (GWO). Initially, the model observed through the comparison of its respective confusion
achieved an impressive accuracy of approximately 99.4% without matrices before and after the incorporation of Grey Wolf
GWO, as depicted in the first matrix. However, upon implementing Optimization (GWO). Initially, the GBC exhibited a modest
GWO, there was a marginal improvement in accuracy, with the accuracy of approximately 79.2456%. However, upon
model's performance rising to around 99.6%, as evidenced in the implementing GWO, a remarkable transformation occurred,
subsequent matrix. elevating the accuracy to an impressive 100%. This stark
improvement underscores the significant impact of GWO in
enhancing the performance of the Gradient Boosting Classifier,
leading to a perfect predictive accuracy.
b) Comparision with other work
c) Comments on Result
The project embarked on an extensive exploration of various Trinidadian Diaspora,” Genealogy, vol. 3, p. 15, Apr.
machine learning algorithms, including SVM, Logistic Regression, 2019, doi: 10.3390/genealogy3020015.
Random Forest Classifier, GBC, and GWO, to predict trends in the A. Derakhshan and H. Beigy, “Sentiment analysis on stock social media for stock
NIFTY50 index. Each algorithm was carefully considered for its price movement prediction,” Eng Appl Artif Intell, vol. 85, pp. 569–578, Oct. 2019,
unique strengths in analyzing financial data, aiming to discern the doi: 10.1016/j.engappai.2019.07.002.
F. Audrino, F. Sigrist, and D. Ballinari, “The impact of sentiment and attention
most effective method for accurate predictions across diverse measures on stock market volatility,” Int J Forecast, vol. 36, no. 2, pp. 334–357,
market conditions. To enrich the predictive models, data from 2020, doi: https://doi.org/10.1016/j.ijforecast.2019.05.010.
financial news platforms and social media channels were S. Mukherjee, B. Sadhukhan, N. Sarkar, D. Roy, and S. De, “Stock market
amalgamated using tools such as Twitter API, Facebook, YouTube, prediction using deep learning algorithms,” CAAI Trans Intell Technol, vol. 8, no. 1,
pp. 82–94, Mar. 2023, doi: 10.1049/cit2.12059.
and Stockwits. Employing sentiment analysis techniques facilitated R. Pandey, A. Mandal, and A. Kumar, “Identifying Applications of Machine
the extraction of valuable insights from textual content sourced Learning and Data Analytics Based Approaches for Optimization of Upstream
from social media platforms, shedding light on investor sentiment Petroleum Operations,” Energy Technology, vol. 8, Jan. 2021, doi:
and market trends. Amidst challenges like the impact of social 10.1002/ente.202000749.
J. Long, Z. Chen, W. He, T. Wu, and J. Ren, “An integrated framework of deep
media, data abundance, and market volatility, the project learning and knowledge graph for prediction of stock price trend: An application in
capitalized on machine learning and sentiment analysis to extract Chinese stock exchange market,” Appl Soft Comput, vol. 91, p. 106205, Mar. 2020,
actionable insights from vast datasets. Techniques like outlier doi: 10.1016/j.asoc.2020.106205.
detection and data standardization were instrumental in ensuring S. Ateş, A. Coskun, M. Sahin, and M. Demircan, “Impact of Financial Literacy on
the Behavioral Biases of Individual Stock Investors: Evidence from Borsa Istanbul,”
the quality of the dataset, thereby mitigating potential errors. Real- Business and Economics Research Journal, vol. 7, p. 1, Sep. 2016, doi:
world data sourced from Yahoo Finance and social media 10.20409/berj.2016321805.
platforms underwent meticulous preprocessing to cleanse and W. Khan, M. Ali Ghazanfar, M. Awais Azam, A. Karami, K. H. Alyoubi, and A. S.
prepare it for model development. Techniques like Z-score Alfakeeh, “Stock market prediction using machine learning classifiers and social
media, news,” J Ambient Intell Humaniz Comput, 2022, [Online]. Available:
normalization and feature engineering further enhanced the quality http://www.finance.yahoo.com
and relevance of the dataset. By seamlessly integrating traditional W. Khan, M. ali Ghazanfar, M. A. Azam, A. Karami, K. Alyoubi, and A. Alfakeeh,
stock analysis with cutting-edge technologies, the project bridges “Stock market prediction using machine learning classifiers and social media,
the gap between established practices and emerging methodologies. news,” J Ambient Intell Humaniz Comput, vol. 13, Jul. 2022, doi: 10.1007/s12652-
020-01839-w.
Through the fusion of machine learning and sentiment analysis, D. Garg and P. Tiwari, “Impact of social media sentiments in stock market
coupled with data from financial news and social media, it delivers predictions: A bibliometric analysis,” Business Information Review, vol. 38, no. 4,
invaluable insights into market dynamics, thereby enhancing pp. 170–182, Dec. 2021, doi: 10.1177/02663821211058666.
prediction accuracy and empowering informed decision-making for A. Jayanth Balaji, D. S. Harish Ram, and B. B. Nair, “Applicability of Deep
Learning Models for Stock Price Forecasting An Empirical Study on BANKEX
investors and financial institutions alike. Data,” Procedia Comput Sci, vol. 143, pp. 947–953, 2018, doi:
https://doi.org/10.1016/j.procs.2018.10.340.
H. CONCLUSIONS T.-W. Sr and C.-C. Sr, “Forecasting Stock Market with Neural Networks,” SSRN
Electronic Journal, Jan. 2009, doi: 10.2139/ssrn.1327544.
Social media as a source of information and sentiment analysis M. Wen, P. Li, L. Zhang, and Y. Chen, “Stock Market Trend Prediction Using
has grown in popularity, especially in the stock market. This is High-Order Information of Time Series,” IEEE Access, vol. PP, p. 1, Feb. 2019, doi:
because investors and traders alike have found that social media 10.1109/ACCESS.2019.2901842.
sentiment can serve as a valuable indicator of market sentiment and L. Ertuna, Stock Market Prediction Using Neural Network Time Series Forecasting.
2016. doi: 10.13140/RG.2.1.1954.1368.
future market trends. This study focused on the Nifty50 index Y. Ding, N. Sun, J. Xu, P. Li, J. Wu, and S. Tang, “Research on Shanghai Stock
between the sample period of 2018 to 2021 and sentiments from an Exchange 50 Index Forecast Based on Deep Learning,” Math Probl Eng, vol. 2022,
investor perspective as the investors are primarily active on such pp. 1–9, Mar. 2022, doi: 10.1155/2022/1367920.
platforms. A robust, high-performance deep learning model has I. H. Isidoros Perikos, “Recognizing emotions in text using ensemble of classifiers,”
been developed to predict the Nifty50 index movement using social Eng Appl Artif Intell, pp. 191–201, 2016.
G. Yang, H. He, and Q. Chen, “Emotion-Semantic Enhanced Neural Network,”
media sentiments and Nifty50 market positions. The proposed
IEEE/ACM Trans Audio Speech Lang Process, vol. PP, p. 1, Dec. 2018, doi:
model consisted of LSTM and DCNN networks whose hyper- 10.1109/TASLP.2018.2885775.
parameters have been carefully tuned, and more than 95% M. Birjali, M. Kasri, and A. Beni-Hssane, “A comprehensive survey on sentiment
prediction accuracies have been achieved. We find that the hybrid analysis: Approaches, challenges and trends,” Knowl Based Syst, vol. 226, p.
approach combining the market positions with social media 107134, 2021, doi: https://doi.org/10.1016/j.knosys.2021.107134.
sentiments yielded an improved prediction performance. Further S. Garg, D. Panwar, A. Gupta, and R. Katarya, “A Literature Review On Sentiment
Analysis Techniques Involving Social Media Platforms,” pp. 254–259, Nov. 2020,
enhancements could make it more accessible, inclusive, and doi: 10.1109/PDGC50313.2020.9315735.
intuitive for diverse individuals, analysts, and policymakers. The D. P. A. G. & R. K. S. Garg, “A literature review on sentiment analysis techniques
proposed model is robust and dependable to forecast NIFTY50 involving social media platforms. ,” in In Sixth International Conference on
returns. In future work, we intend to investigate the performance of Parallel, Distributed and Grid Computing , 2020, pp. 254–259.
the proposed model to examine the returns of the sector-specific N. Singh, Sentiment Analysis on Motor Vehicles Amendment Act, 2019 an Initiative
by Government of India to follow traffic rule. 2020. doi:
market indices. 10.1109/ICCCI48352.2020.9104207.
B. Weng, L. Lu, X. Wang, F. Megahed, and W. Martinez, “Predicting Short-Term
REFERENCES Stock Prices using Ensemble Methods and Online Data Sources,” Expert Syst Appl,
vol. 112, Jun. 2018, doi: 10.1016/j.eswa.2018.06.016.
D. Plaza and L. Plaza, “Facebook and WhatsApp as L. L. X. W. F. M. W. G. M. Bin Weng, “Predicting short-term stock prices using
Elements in Transnational Care Chains for the ensemble methods and online data sources,” Expert Syst Appl, Dec. 2018.
D. E. Farrar and R. R. Glauber, “Multicollinearity in Regression Analysis: The Programs Biomed, vol. 213, p. 106504, 2022, doi:
Problem Revisited,” Rev Econ Stat, vol. 49, no. 1, pp. 92–107, 1967, doi: https://doi.org/10.1016/j.cmpb.2021.106504.
10.2307/1937887. N. Baracaldo, B. Chen, H. Ludwig, A. Safavi, and R. Zhang, Detecting Poisoning
C. Dormann et al., “Collinearity: A review of methods to deal with it and a Attacks on Machine Learning in IoT Environments. 2018. doi:
simulation study evaluating their performance,” Ecography, vol. 36, pp. 27–46, Apr. 10.1109/ICIOT.2018.00015.
2013, doi: 10.1111/j.1600-0587.2012.07348.x. L. Breiman, “Random Forests,” Mach Learn, vol. 45, no. 1, pp. 5–32, 2001, doi:
H. C. Mandhare and S. R. Idate, “A comparative study of cluster based outlier 10.1023/A:1010933404324.
detection, distance based outlier detection and density based outlier detection S. Suthaharan, “Chapter 6 - A Cognitive Random Forest: An Intra- and
techniques,” in 2017 International Conference on Intelligent Computing and Intercognitive Computing for Big Data Classification Under Cune Condition,” in
Control Systems (ICICCS), 2017, pp. 931–935. doi: Handbook of Statistics, vol. 35, V. N. Gudivada, V. V Raghavan, V. Govindaraju,
10.1109/ICCONS.2017.8250601. and C. R. Rao, Eds., Elsevier, 2016, pp. 207–227. doi:
B. Wang, G. Xiao, H. Yu, and X. Yang, “Distance-Based Outlier Detection on https://doi.org/10.1016/bs.host.2016.07.006.
Uncertain Data,” in 2009 Ninth IEEE International Conference on Computer and Y. Mishina, R. Murata, Y. Yamauchi, T. Yamashita, and H. Fujiyoshi, “Boosted
Information Technology, 2009, pp. 293–298. doi: 10.1109/CIT.2009.107. Random Forest,” IEICE Trans Inf Syst, vol. E98.D, pp. 1630–1636, Sep. 2015, doi:
V. Sharma, “A Study on Data Scaling Methods for Machine Learning,” 10.1587/transinf.2014OPP0004.
International Journal for Global Academic & Scientific Research, vol. 1, Feb. 2022, J. Ali, R. Khan, N. Ahmad, and I. Maqsood, “Random Forests and Decision Trees,”
doi: 10.55938/ijgasr.v1i1.4. International Journal of Computer Science Issues(IJCSI), vol. 9, Sep. 2012.
D. U. Ozsahin, M. T. Mustapha, A. S. Mubarak, Z. S. Ameen, and B. Uzun, “Impact Y.-Y. Song and Y. Lu, “Decision tree methods: applications for classification and
of feature scaling on machine learning models for the diagnosis of diabetes,” in prediction,” Shanghai Arch Psychiatry, vol. 27, pp. 130–135, Apr. 2015, doi:
2022 International Conference on Artificial Intelligence in Everything (AIE), 2022, 10.11919/j.issn.1002-0829.215044.
pp. 87–94. doi: 10.1109/AIE57029.2022.00024. P. Probst, M. N. Wright, and A. Boulesteix, “Hyperparameters and tuning strategies
R. Kumar Pandey, A. Gandomkar, B. Vaferi, A. Kumar, and F. Torabi, “Supervised for random forest,” Wiley Interdiscip Rev Data Min Knowl Discov, vol. 9, 2018,
deep learning-based paradigm to screen the enhanced oil recovery scenarios,” Sci [Online]. Available: https://api.semanticscholar.org/CorpusID:4753950
Rep, vol. 13, no. 1, p. 4892, 2023, doi: 10.1038/s41598-023-32187-2. M. D. Guillen, J. Aparicio, and M. Esteve, “Gradient tree boosting and the
I. Parmar et al., “Stock Market Prediction Using Machine Learning,” in 2018 First estimation of production frontiers,” Expert Syst Appl, vol. 214, p. 119134, 2023, doi:
International Conference on Secure Cyber Computing and Communication https://doi.org/10.1016/j.eswa.2022.119134.
(ICSCCC), 2018, pp. 574–576. doi: 10.1109/ICSCCC.2018.8703332. K. F. Hew, X. Hu, C. Qiao, and Y. Tang, “What predicts student satisfaction with
T. Evgeniou and M. Pontil, Support Vector Machines: Theory and Applications, vol. MOOCs: A gradient boosting trees supervised machine learning and sentiment
2049. 2001. doi: 10.1007/3-540-44673-7_12. analysis approach,” Comput Educ, vol. 145, p. 103724, 2020, doi:
M. Baldomero-Naranjo, L. I. Martínez-Merino, and A. M. Rodríguez-Chía, “A https://doi.org/10.1016/j.compedu.2019.103724.
robust SVM-based approach with feature selection and outliers detection for A. Natekin and A. Knoll, “Gradient Boosting Machines, A Tutorial,” Front
classification problems,” Expert Syst Appl, vol. 178, p. 115017, 2021, doi: Neurorobot, vol. 7, p. 21, Dec. 2013, doi: 10.3389/fnbot.2013.00021.
https://doi.org/10.1016/j.eswa.2021.115017. T. Chen and C. Guestrin, XGBoost: A Scalable Tree Boosting System. 2016. doi:
S. Ghosh, A. Dasgupta, and A. Swetapadma, A Study on Support Vector Machine 10.1145/2939672.2939785.
based Linear and Non-Linear Pattern Classification. 2019. doi: M. Machado, S. Karray, and I. Sousa, LightGBM: an Effective Decision Tree
10.1109/ISS1.2019.8908018. Gradient Boosting Method to Predict Customer Loyalty in the Finance Industry.
M. Maalouf, “Logistic regression in data analysis: An overview,” International 2019. doi: 10.1109/ICCSE.2019.8845529.
Journal of Data Analysis Techniques and Strategies, vol. 3, pp. 281–299, Jul. 2011, S. Mirjalili, S. Mirjalili, and A. Lewis, “Grey Wolf Optimizer,” Advances in
doi: 10.1504/IJDATS.2011.041335. Engineering Software, vol. 69, pp. 46–61, Mar. 2014, doi:
J. Peng, K. Lee, and G. Ingersoll, “An Introduction to Logistic Regression Analysis 10.1016/j.advengsoft.2013.12.007.
and Reporting,” Journal of Educational Research - J EDUC RES, vol. 96, pp. 3–14, Z. Yang, “Competing leaders grey wolf optimizer and its application for training
Sep. 2002, doi: 10.1080/00220670209598786. multi-layer perceptron classifier,” Expert Syst Appl, vol. 239, p. 122349, 2024, doi:
A. Bailly et al., “Effects of dataset size and interactions on the prediction https://doi.org/10.1016/j.eswa.2023.122349.
performance of logistic regression and deep learning models,” Comput Methods