Toward Smart Bicycle Safety: Leveraging Machine Learning Models and Optimal Lighting Solutions

Hence the need to minimize injuries resulting from bicycle accidents, and even though cycling leads to befitting health as well environmental influence since cyclist are far much vulnerable embodied on road makes reducing harm out of bicycling crashes by all ways. But these collisions can still happen, the role of bicycle lighting being one of such factors and contributing to their quality. In the meantime, the rise of e-commerce has created a tidal wave of user-generated content in this space alone that gained popularity for providing product reviews. Customer feedback is an important element in Amazon, which is a leading Internet marketplace. This becomes a challenge when conducting analysis, which has led to the development of sophisticated systems that classify them according to their types. This paper gives a summary of the work that has been conducted in research on categorization of Amazon reviews concerning lights used for bikes. The main goal is to group the reviews as positive and negative opinions which help for business organizations or consumers gives a brief idea of what people say about that particular product. Many different machine learning approaches and natural language processing methods have been used for effective solving this task. At first, we started with the extraction of pertinent textual attributes and used several supervised learning algorithms such as J48 classifier (J)1 to logistic regression (LR), Naive Bayes (NB), Stochastic Gradient Descent classifier SGD, and K-Nearest Neighbor (KNN). We generated models through labeled review training datasets. Then compared the proposed models and got the following accuracy: 94, 82, 92, and 94%. From the above findings, it is clear that the LR and KNN models performed better than their counterparts thus can be considered as such effective in addressing this task.