suggested cost sensitive strategy may successfully lower the Xianghui Yuan et al.[9] analyses the financial stability of price of international investments and increase profitability. several combined choices of stock models that contain At last, this study gives us knowledge on the usage of various feature selection and stock price trend prediction machine learning models for stock market forecasting and algorithms. They construct such models using methods emphasises the need to take into account cost sensitive including feature selection, time-sliding windows for cross applications in this field. validation, and supervised methods. When choosing the top 1% of equities, the RF-RF integrated model occupies a place Mahinda Mailagaha Kumbure et al.[4] highlighted that of profitable, producing a 29.51% annualised return. With an previous evaluations frequently focused on library analysis, annualised return of 21.92% and a manageable maximum finding highly cited publications and individuals, while drawdown of 13.58% from 2011 to 2018, further giving few aspects about the estimation techniques, input examination utilising a divided back- testing approach factors, and data used in this research. The current work was supports the model's profitability. motivated by the lack of knowledge to offer a thorough analysis of machine learning models that are usually used in Jui-sheng chou et al.[10] explains how to preview stock financial forecasting. In-depth analyses of forecasting prices using support vector regression, also known as SVR, methodologies, the subtleties of input variables and data and its variation, least squares support vector regression. sources, and a close examination of particular variables and LSSVR is renowned for depending on hyperparameter variable types utilised in the speaking of market forecasts are adjustment for optimum performance and is praised for its all covered in the study, that offers a better view than earlier effectiveness in addressing massive regression issues. evaluations. Artificial bee colonies, cuckoo searches, particle swarm optimisation, and the firefly optimisation method are just a Mohamed Sharaf et al.[5] tells that it is a habit for few examples of nature inspired optimisation methods that researchers to analyse earlier literature to expand on current are being involved to automate the hyperparameter knowledge and spot research gaps that influence their work. optimisation process. Particular values are placed on the These connected studies may include conventional theories firefly algorithm (FA) due to its effectiveness in resolving of portfolios and number of Machine Learning models for numerous optimisation issues, including market price stock prediction, and also earlier attempts to optimise predictions. Fundamentals and their variations are referred to portfolios using various methods and techniques. be effective tools for design and engineering optimisation Researchers can place their investigation within the larger jobs. framework of earlier research and highlight the originality and contributions of their suggested strategy by looking at Sharanya Banerjee et al.[11] discusses several techniques and citing comparable work. that are evaluated on technical analysis such as simple linear regression models, support vector machines with Principal YOUNGBIN KIM et al.[6] assumes that predicting stock Component Analysis (PCA), Artificial Neural Networks prices includes using a variety of data sources and (ANN) with sentiment analysis, Hidden Markov Models, algorithms. Researchers use text-based data from news and hybrid models using sentiment analysis and clustering social media including numerical elements from previous methods, LSTM (Long Short-Term Memory) networks, and market data. For time-series modelling, they frequently LSTM combined with clustering methods. Each of these employ Recurrent Neural Networks (RNNs) such as LSTM algorithms aims to anticipate stock prices.ck prices and and GRU, to identify sequential trends. Several models trends. investigate the impact of business events on stock prices while others include sentiment analysis to identify investor KYUNGBOK MIN et al.[12] explains how to utilise stock feelings from text. Transformer encoders capture temporal fundamentals to anticipate stock price trends by analysing correlations inside and between stocks, whereas graph-based the correlation between financial news stories and historical systems describe inter-company interactions. price movements in stocks. Although earlier studies demonstrated the efficacy of such methods, they frequently NAGARAJ NAIK et al.[7] describes correlations between ignored the structural links inside phrases. Due to its the shares and currency rates, the usage of Support Vector capacity to preserve memory across numerous time steps, Machines (SVM) and Artificial Neural Networks (ANN), recurrent neural networks are presented to be most effective specifically in a setting of Korean stock market information, alternative to convolutional neural networks (CNNs) for text and the application of Autoregressive Integrated Moving and audio processing. However, when it comes to Average (ARIMA) models alone or in combination to dependency over time, RNNs experience the issue of capture non-linear patterns. The article also discusses the gradients disappearing. long short term memory and a model importance of Deep Neural Networks (DNN) which will known as Gated Recurrent Unit (GRU) units are suggested validate the scant literature on equity crisis prediction, as remedies to this problem, with GRU being statistically giving ideas for extended research in this field. The paper more effective. Schuster's model is where the concept of a offers insightful analyses of the various price forecasting two-stream GRU was first introduced, and it is also techniques and their applications. introduced in the context which demonstrates how machine Daiyou Xiao et al.[8] offers an overview of how Machine learning algorithms are utilised for predicting stock price Learning Models are being used within the banking industry, trends by examining the relationship among financial news particularly to forecast stock values using both conventional items and past stock price changes. Even if past research and machine learning models. The study depicts the case showed that these techniques worked, they typically study of a New York Stock Exchange dataset that spans the overlooked the structural links inside phrases. years 2010 to 2019. To predict stock prices and their HAMED JABANI et al.[13]. gives information about how to correlations among them the two basic models such as properly estimate stock trends using a combination of ARIMA (autoregressive integrated moving average) and algorithms and methodologies. Firstly, it utilises different LSTM (long short-term memory) neural networks are used. economic indicators, including the Simple Moving Average The outcome of experimentation shows that all of the (SMA), Weighted Moving Average (WMA), Momentum models have the ability of accurately forecasting correlations (MOM), Stochastic Oscillator (STCK), Stochastic Oscillator and shares, with LSTM beating ARIMA in this regard. Signal Line (STCD), Larry Williams Percent Range (LWR), VOLUME 18, ISSUE 4, 2024 32 http://xadzkjdx.cn/
