A Machine Learning Framework for Stock Prediction using Sentiment Analysis

Sentiment analysis, which stands for opinion mining, is a natural language processing (NLP) technique that involves identifying, extracting, and analyzing sentiments or opinions expressed in text data. The primary goal of sentiment analysis is to determine the sentiment polarity of a given piece of text, whether it is positive, negative, or neutral. This analysis can be applied to various types of content, such as product reviews, social media posts, customer feedback, and news articles. Sentiment analysis algorithms use machine learning and text classification to understand subjective information conveyed in text, helping businesses, organizations, and individuals gain insight into public opinions and emotions about specific topics, products, or services. In this study, we conducted sentiment analysis on a Bengali dataset. For feature extraction, we implemented the term frequency-inverse document frequency (TF-IDF) technique, and for feature selection, we applied an extra tree classifier approach. Subsequently, we trained our machine learning model, achieving an impressive accuracy rate of 92%.

Machine Learning (ML) is an Artificial Intelligence (AI) approach that allows systems to adapt to their environment based on past experiences. Machine Learning (ML) and Natural Language Processing (NLP) techniques are commonly used in sentiment analysis and Information Retrieval Techniques (IRT). This study supports the use of ML approaches, such as K-Means, to produce accurate outcomes in clustering and classification approaches. The main objective of this research is to explore the methods for sentiment classification and Information Retrieval Techniques (IRT). So, a combination of different machine learning algorithms is used with a dataset from amazon unlocked mobile reviews and telecom tweets to achieve better accuracy as it is crucial to consider the previous predictions related to sentiment classification and IRT. The datasets consist of user reviews ratings and algorithms utilized consist of K-Means Clustering algorithm, Logistic Regression (LR), Random Forest (RF), and Decision Tree (DT) algorithms. The amalgamation of each algorithm with the K-Means resulted in high levels of accuracy. Specifically, the K-Means combined with Logistic Regression (LR) yielded an accuracy rate of 99.98%. Similarly, the K-Means integrated with Random Forest (RF) resulted in an accuracy of 99.906%. Lastly, when the K-Means was merged with the Decision Tree (DT) Algorithm, the accuracy obtained was 99.83%.We exhibited that we could foresee efficient, effective, and accurate outcomes.

Social media has dramatically influenced and changed the rate and the nature of crime in our society. The perpetrators cut across different age groups, social standing, and beliefs. The ability to be anonymous on social media and the lack of adequate resources to fight cybercrime are catalysts for the rise in criminal activities, especially in South Africa. We proposed a system that will analyse and detect crime in social media posts or messages. The new system can detect attacks and drug-related crime messages, hate speech, and offensive messages. Natural language processing algorithms were used for text tokenisation, stemming, and lemmatisation. Machine learning models such as support vector machines and random forest classifiers were used to classify texts. Using the support vector machine to detect crime in texts, we achieved 86% accuracy and using the random forest for crime analysis, 72% accuracy was achieved.KeywordsCrime detectionSocial mediaNatural language processingSupport vector machineRandom forest

On the relationship between changes in stock prices and bond yields in the G7 countries: Wavelet analysis

Although the prediction of stock prices is a topic that has been studied in several occasions, there are few variables that have been used for this matter outside of dividends such as determinant of prices (Guo, 2022). The results found to date have not been entirely satisfactory. since they do not explain most price movements in the long run. The use of profits such as predictor variable is an alternative that has not been much explored. This research work seeks to answer what is the effect of using profits as a predictor of stock prices in the long run. With the In order to answer this question, a panel regression was performed, controlling for fixed effects, of the effect of the profits on stock prices over the past 30 years for S&P 500 companies. In general, it was not possible to determine if the profits, whether gross, operating or net, have the capacity to predict changes in stock prices. However, it was established that the size of the assets of a company and changes in US GDP have a negative relationship with changes in price. By On the other hand, the change in the price of the SPX index has a positive relationship with the change in the stock price. This finding can generate a great advantage for investors when creating stock portfolios in the long term that seek to maximize profitability with a certain degree of certainty, since it would allow choosing with greater clarity those actions with greater probability of success and returns in the long

Now-a-days, Social media platforms like Facebook, Twitter etc. are becoming more and more popular to express feelings and thoughts. People not only share their happy moments on these platforms but also share their feelings when they are extremely depressed. Analyzing these social media posts, one’s mental condition can be detected whether he/she is happy, sad or angry at a particular time using sentiment analysis in natural language processing. Most of the research in this field is based on English language and the accuracy of sentiment analysis from Bengali language is not very high. So, our purpose is to work on this field using Bengali dataset collected from different social media posts and make the sentiment detection more accurate so that this work can be used to build a system that can be used in the mental health sector of our country. In this research, we have first collected social media data. After applying different data preprocessing techniques, we have made a number of feature selection and extraction combinations. We have applied some basic and advanced machine learning and deep learning algorithms on each of these to find the best combination and algorithm by monitoring the highest achieved accuracy.

There is a vast amount of unstructured text data generated every day analyzing and making sense of these text-based datasets is a complex, cumbersome task. The existing visualization tools that analyze text data leveraging Natural Language Processing (NLP) techniques, are often tailored for structured text-based data. They also fail to support reading, a crucial analysis task to validate the output of NLP techniques. We designed and developed TextVista, an NLP-enriched visualization tool that supports analysts during their analysis of unstructured text with temporal references. Our tool combines techniques including clustering, sentiment analysis, and threat detection with three views that visualize high-level patterns in the data to encourage reading. We report on TextVista’s iterative design process, which included a focus group to distill design requirements, a think-aloud interview study with data analysts to understand their impressions of the tool, and a diary study to assess its long-term usage. Through this process, we identified how TextVista supported the analysis of unstructured text with temporal references using NLP techniques and fostered methods to promote reading in situ. TextVista also encouraged serendipity when analyzing data via its question-focused overviews and flexible avenues to explore data.

Since the two last decades, social media networks have become a part of our daily life. Today, getting information from social media, tracking trends in social media, learning the feelings and emotions of people on social media is very essential. In this study, sentiment analysis was performed on Twitter text to learn about the subjective polarities of the writings. The polarities are positive, negative, and neutral. At the first stage of the sentiment analysis, a public data set has been obtained. Secondly, natural language processing techniques have been applied to make the data ready for machine learning training procedures. Lastly, sentiment analysis is performed by using three different machine learning algorithms. We reached 89% accuracy with Support Vector Machines, 88% accuracy with Random Forest, and 72% accuracy with Gaussian Naive Bayes classifier.

BackgroundIdentifying patients at high risk of stroke-associated pneumonia (SAP) may permit targeting potential interventions to reduce its incidence. We aimed to explore the functionality of machine learning (ML) and natural language processing techniques on structured data and unstructured clinical text to predict SAP by comparing it to conventional risk scores.MethodsLinked data between a hospital stroke registry and a deidentified research-based database including electronic health records and administrative claims data was used. Natural language processing was applied to extract textual features from clinical notes. The random forest algorithm was used to build ML models. The predictive performance of ML models was compared with the A2DS2, ISAN, PNA, and ACDD4 scores using the area under the receiver operating characteristic curve (AUC).ResultsAmong 5,913 acute stroke patients hospitalized between Oct 2010 and Sep 2021, 450 (7.6%) developed SAP within the first 7 days after stroke onset. The ML model based on both textual features and structured variables had the highest AUC [0.840, 95% confidence interval (CI) 0.806–0.875], significantly higher than those of the ML model based on structured variables alone (0.828, 95% CI 0.793–0.863, P = 0.040), ACDD4 (0.807, 95% CI 0.766–0.849, P = 0.041), A2DS2 (0.803, 95% CI 0.762–0.845, P = 0.013), ISAN (0.795, 95% CI 0.752–0.837, P = 0.009), and PNA (0.778, 95% CI 0.735–0.822, P < 0.001). All models demonstrated adequate calibration except for the A2DS2 score.ConclusionsThe ML model based on both textural features and structured variables performed better than conventional risk scores in predicting SAP. The workflow used to generate ML prediction models can be disseminated for local adaptation by individual healthcare organizations.

A company's financial statements provide information that is useful in making economic decisions. This may include information about the company's financial performance, financial position and cash flows. Valuation of financial instruments at fair value can have a significant impact on bank share prices. Banks that use fair value accounting for financial instruments may experience greater variability in their share prices than banks that do not use this method. The purpose of this research is to investigate the relationship between accounting valuation of financial instruments at fair value and changes in bank stock prices. The research aims to determine whether and to what extent there is a relationship between the fair value of financial instruments and changes in stock prices. Research may also aim to identify any factors that may influence this relationship and provide information and recommendations for stakeholders such as investors, regulators and financial institutions. Based on the results achieved, the need for optimal and accurate use of fair value indicators due to their impact on the company's performance, as well as clearer compliance by banks with international financial reporting standards regarding the fair value assessment of financial instruments, was investigated. And thus, it can lead to increased accuracy of forecasts and increased confidence of users of financial statements in making economic decisions. The findings of the study can contribute to the existing scientific literature on the impact of fair value accounting on stock prices and provide further understanding of this issue. The results of the research can be used by banks to assess the potential impact of the implementation of fair value accounting for financial instruments on the share price.

Six text classification methods were compared to find the best model for predicting Stock Exchange of Thailand stock prices. News headlines, on individual stocks, were classified as causing “change” and “no-change” based on a preset change threshold, 2.5%. The training dataset was collected by matching stock news in 2018 with stock names and filling in stock price changes. 258 news were associated with a “change” and 636 news with “no-change”. The Thai text news items were preprocessed and converted to TF-IDF vector representation. Six machine learning text classification methods are applied to create six text classifier models and create a confusion matrix, then compared with actual changes to obtain accuracy scores. We found that a deep learning classifier (with 85.6% accuracy) scored better than other classifiers for one day price movement to assist short-term investments.

In recent years, ride-hailing platforms such as Uber have significantly transformed the transportation industry by offering convenient, flexible, and technology-driven services. These platforms have made travel more accessible by enabling users to book rides instantly through mobile applications. However, customer satisfaction is not always consistent, as users frequently report issues related to pricing, driver behavior, service reliability, and application performance. Understanding these concerns is essential for improving service quality and enhancing user experience.This study focuses on analyzing customer reviews collected from Uber using sentiment analysis techniques. Since customer feedback is typically unstructured and large in volume,manual analysis is inefficient. Therefore, Natural Language Processing (NLP) techniques are used to process and interpret the textual data. The research applies both machine learning and deep learning models, including Logistic Regression, Naïve Bayes, Random Forest, and Long Short-Term Memory (LSTM), to classify reviews into positive, negative, and neutral sentiments. The dataset used in this study consists of approximately 12,000 customer reviews, which provide a diverse representation of user experiences. Data preprocessing techniques such as cleaning, tokenization, stopword removal, and stemming are applied to improve data quality. The experimental results demonstrate that deep learning models, particularly LSTM,outperform traditional machine learning approaches in terms of accuracy and contextual understanding. The proposed model achieved an overall accuracy of 90%, indicating its effectiveness in correctly classifying customer sentiments. The analysis also reveals key factors influencing customer satisfaction, including frequent ride cancellations, surge pricing, driver-related issues, and technical problems within the application. Overall, this research highlights the importance of sentiment analysis as a powerful tool for understanding customer opinions in ride-hailing platforms. The insights obtained from this study can help companies make data-driven decisions, improve service quality, and enhance customer satisfaction.

This Chapter discusses the application of Natural Language Processing (NLP) in the legal domain for entity identification and the generation of hierarchy-based applications. NLP techniques, such as Named Entity Recognition (NER) and information extraction, are utilized to analyse legal documents, extract key information, and understand contextual relationships. The integration of NLP facilitates tasks such as legal document summarization, contract analysis, automated document generation, and legal research, thereby enhancing the efficiency and accuracy of legal processes. The creation of hierarchical structures within documents, language translation, sentiment analysis, and compliance monitoring further contribute to the comprehensive utilization of NLP in the legal field. The abstract highlights the transformative impact of NLP in streamlining legal workflows, improving information retrieval, and ensuring compliance with evolving legal standards. Most judicial systems in the world, centered around the Supreme Court as the highest-level court, grapples with a substantial caseload, receiving over 7,500 cases annually but hearing fewer than 150. The Supreme Court's unique role in interpreting the constitutionality of laws and establishing precedents underscores its significant influence. Challenges arise from the overwhelming caseload, leading to increased use of plea bargains and potential concerns about justice. To address these issues, a proposed research initiative suggests employing machine learning and natural language processing to accelerate Supreme Court decisions, aiming to uncover pivotal aspects that greatly influence the outcomes, and enhance our comprehension on India legal system's functioning and constraints. This chapter presents a methodology for sentiment-based summarization of legal documents using Natural Language Processing (NLP) techniques. The process involves pre-processing the text, conducting sentiment analysis to label sentences as positive, negative, or neutral, and then summarizing the document while incorporating the identified sentiments. Special attention is paid to legal-specific considerations such as terminology and context.

The rapid growth of social media platforms has significantly influenced human behavior, emotions, and mental well-being. With millions of users expressing their thoughts and feelings online, social media has become a valuable source of psychological insights. This project proposes an AI-based Psychology Analysis and Depression Detection System that leverages Machine Learning (ML) and Natural Language Processing (NLP) techniques to identify early signs of depression from social media content. The system collects publicly available textual data such as posts, comments, and user interactions, and applies advanced text preprocessing methods including tokenization, sentiment analysis, and feature extraction. Machine learning models such as Logistic Regression, Support Vector Machines (SVM), and Deep Learning approaches like LSTM are trained to classify user mental states based on linguistic patterns, emotional tone, and behavioral indicators. Additionally, the project analyzes user engagement patterns, posting frequency, and sentiment trends over time to enhance prediction accuracy. The proposed model aims to provide early detection and psychological insights, enabling timely intervention and support for individuals at risk of depression. This AI-driven approach can assist mental health professionals, researchers, and organizations in understanding online psychological behavior while ensuring ethical data handling and user privacy protection.

This paper considers variance bounds for stock price changes in a general setting that allows for ex-dividend stock prices, risk averse investors, and exponentially-growing dividends. I show that providing investors with more information about future dividends can either increase or decrease the variance of stock price changes, depending on some key parameters, namely, those governing the properties of dividends and the stochastic discount factor. This finding contrasts with the results of Engel (2005) who shows that news about future dividends will always decrease the variance of stock price changes in a specialized setting with cum-dividend stock prices and risk neutral investors.