International Journal of Data Science and Advanced Analytics

Air Quality Transformation in Twelve Major Cities during Covid-19 Lockdowns: A Global Assessment

Ismail I Abbas — 2023-11-24

The implementation of lockdown measures worldwide, aimed at preventing the spread of Covid-19, has temporarily improved air quality. This research paper aims to examine the impact of the lockdown period from March to May 2020 on the levels of four common air pollutants in 12 major cities, namely Delhi (India), Newcastle (UK), California (USA), Brescia (Italy), São Paulo (Brazil), Langfang (China), Madrid (Spain), Khon Kaen (Thailand), Santiago (Chile), Bogota (Colombia), Wellington (New Zealand), and Silivri (Turkey). The study analyzed the changes in average monthly concentrations of nitrogen dioxide (NO₂), ozone (O₃), and particulate matter (PM10 and PM_2.5) during two phases: the pre-lockdown and lockdown phases. During the lockdown, all air pollutants except ozone exhibited a significant decrease. These results highlight the positive impact of reducing anthropogenic emissions on air quality during the Covid-19 lockdown. The researchers also used principal components analysis to examine the concentrations of NO₂, O₃, PM_2.5, and PM10 from January 2018 to June 2020. The findings revealed that the 11 monitoring sites in the cities could be grouped into six clusters based on similar air pollution patterns. Overall, this study provides valuable insights that can inspire policymakers and stakeholders involved in air quality management to implement changes in environmental policies. By targeting pollution sources, it is possible to mitigate the harmful effects of air pollutants.

Probability Graphical Model for Predicting Probability of Default For Mortgage Loans

Trupti Wagh — 2023-11-13

Assessment of Default risk of borrowers is important for lending institutions as it directly affects profits and losses of the firm and guides in compensating the risk by taking appropriate majors for loans having higher probability of default. Predicting probability of default using statistical and machine learning models has been a popular research topic in data science community. While different types of classification models have been proposed historically, there is scope to apply probabilistic inference to the mortgage default analysis to support decision making. Probabilistic Graphical Model (PGM) are a powerful framework for compactly encoding probability distributions over complex multivariate domains using graphical representations. Due to the high interpretability and inherent support for probabilistic inference, the PGM models have widely been used under various domains such as medical diagnosis, text, audio, video processing.

Using Supervised Machine Learning Models and Natural Language Processing for Identification of Fake News

Sidarth Mohan — 2023-11-09

Social media has gained popularity over the last decade due to its ease of access and providing large amount of information to people. In seconds, users are able to access information from social media related to politics, life-style, science and money other fields. However, data obtained from social media platforms represent a mixture of fake and real news. Fake news are in-tended to deceive people and change their attitudes and beliefs. Machine learning algorithms have shown successful in classifying real from fake news. Nonetheless when applying machine learning models in this context related to limitations in the dataset type, balance or skewness. Hence, data pre-processing is essential prior to application of machine learning models. Therefore, this work evaluated the use of supervised machine learning models with different data pre-processing approaches for classification of fake news obtained from social media platforms. Different pre-processing techniques have been applied related to feature extraction and feature selection alongside four machine learning models being logistic regression, decision trees, random forest and extreme gradient boost. The findings showed that random forest and extreme gradient boost with bi-gram feature extraction and chi-squared feature selection showed the best performance. Future work involves using the proposed model to detecting fake news in different con-text and different languages.

Prediction of Customer Sentiment Based on Online Reviews Using Machine Learning Algorithms

Chinmayee Guru — 2023-11-29

Customer opinions and feedback play a pivotal role in enhancing business operations and decision-making processes. Sentiment analysis is a crucial technique used to decipher customer opinions from their feedback and thus provide valuable insights for businesses. However, analysing and understanding reviews is an intricate process and prone to be misleading if not conducted meticulously. This study aims to extract and classify customer emotions from e-commerce reviews of women’s clothing in terms of polarity of sentiment, enhancing sentiment analysis accuracy by means of machine learning (ML) classifiers. In addition, the study addresses the challenge of imbalanced data samples. Several supervised ML models, including Naïve Bayes, Random Forest, Support Vector Classifier (SVC) and Extreme Gradient Boosting (XGB), were employed for sentiment analysis. The study also attempts to deal with negations, and pre-processing steps were implemented to reduce the dimensionality and noise of the raw text. In SVC and XGB models, negation handling significantly improved the precision and recall values of minority classes. Moreover, a hybrid class balancing approach and the synthetic minority oversampling technique (SMOTE) were adopted. The findings indicate that the XGB classifier, combined with SMOTE sampling, produced the most accurate results, yielding better F1 and ROC AUC scores.

Agrifood Firms Collaboration through Brand Orientation Behaviour

Isaac Ngugi — 2023-11-15

This study applies signalling theory as a framework for understanding the pathways between brand orientation behaviour and supply chain collaboration. The path analytic findings based on data obtained from 343 UK-based small- and medium-sized firms in the agri-food industry suggest that a firm’s brand orientation behaviour allows it to improve its level of collaboration with its channel members, and medicating mechanism involved positive network identity, low behaviour uncertainty and social capital. These findings extend the theoretical insights on the relational view, transaction cost economics and strategic network by illustrating how signalling theory can be used to explain the impact of brand orientation behaviour on supply chain collaboration.