Understanding Climate Change Concern Using Machine Learning Technique

Abstract

Climate transformation has been identified as the principal environmental crisis of the 21st century and become a matter of considerable debate. Nowadays, the significance of climate change is extensively argued through various papers, newspapers, websites, and blogs. But that writing may lack accuracy, while the severity of results in scientific articles may be too unclear for the public to understand. With the rapid evolution of the internet and communication technology today, more and more people are interested to express their thoughts via social media and debate there a lot. Utilize the welfare of social media in this study, a gigantic dataset of tweets containing particular keywords connecting to climate change is analyzed using volume analysis and text mining methods such as case modeling and sentiment analysis and how ML can be a strong tool to predict accurately people's concerns and help society adapt to a changing climate. This study provides an explanation as well as a solution by classifying the tweet's natural, man-made, and neutral opinion reviews using the Random Forest (RF), Support Vector Machine (SVM), K-Nearest Neighbor, and Logistic Regression (LR). Among these algorithms, Support Vector Machine accuracy is the best accuracy and accuracy is 83.01%. The proposed model is made on Jupyter Notebook (a Python-based IDE) and trained on Kaggle's standard Twitter Climate Change Sentiment Dataset which has 43,743 records.