Advances in Science and Technology Vol. 124

Abstract: The increase in new cars and customers' economic inability, global sales of old cars are expanding. As a result, there exists a pressing need for a second-hand automobile method for predicting prices that accurately calculates the value of a car based on number of factors. In the current circumstance, the existing system involves a mechanism in which a seller sets a price at random and the buyer has no knowledge of the car or its value. In fact, the seller doesn't even know the current value of the car or the price at which the car should be sold. To solve this problem, we have developed a very effective model. Regression algorithms are used to provide continuous values as output rather than classified values. This allows for the prediction of the car's real price rather than its price range.

Abstract: Pests are the biggest problems faced by farmers. The main objective of Pest classification with deep learning and ReactJS paper is to create a website that takes in input in the form of images from the user and tries to classify them into various pests (mainly 9 different classes). For the classification of pests, three deep learning models were chosen and trained. Their performance is compared and the best performing model is deployed in a form of a website with a good and intuitive user interface. The user interface was created with React (javascript framework), Sass, and TensorFowJS (A deep learning library designed especially for javascript developers). The deep learning models that are selected, trained, and evaluated in this paper are VGGnet-16, ResNet-152, and MobileNet. MobileNet has provided the highest accuracy of 99.80%.

Abstract: Covid-19 is a dangerous infection which is caused by SAR-COV 2 (stands for severe acute respiratory syndrome coronavirus 2). The first known coronavirus case was found in Wuhan, China in 2019. On 11th March 2020 the World health organization declared covid-19 as an epidemic and from that point of time it has been mutated into various variants and thus causing a lot of health issues among the people. It has affected the people as well as the economy of the country in a negative way. Major symptoms which were recognize from the people who were affected from the covid includes cough, fatigue, constant fever, difficulty during breathing and loss of smell and taste. In this study, an overview of analysis of the variants of the coronavirus as well as impact of covid-19 in various countries has been studied in a detailed manner. In order to show the impact of Covid-19 on a clear picture, the study has been visualized with the help of various charts and figures in order to make it easy to understand. Techniques which provide effective visualization is discussed. A comparison of the prediction models with deep learning algorithms is also mentioned in this paper. This study would be helpful for covid-19 research as it gives a clear view of trends and different aspects of covid-19 as well as a detailed view on the variants which were formed.

Abstract: Usually, people really need advice in deciding their choices. Whether it's the movie to watch on Friday night or there are some exciting new things available on e-business. In this distinctive situation, we tend to build a framework which will thus bring down new melodies to shoppers supported their previous history of listening. Now many companies today use the recommendation systems to their advantage such as flip kart and Amazon for the sale of products (e-commerce), ganna.com and music for streaming music, for the sale of clothes, for the films. It assists each business and customers as businesses get financial advantages by attracting customers and users pick up services. Nowadays, everyone uses recommender systems in various forms and they are getting better and better day by day because researchers are trying to cause them to higher and higher each day because of the sturdy competitiveness of the marketplace to provide higher and higher offerings and entice peoples. This project mainly focuses on music only for the music lovers to help them listen to songs they might love. This project allows customers to find new collections or the songs by making the lovey the list accessible for the tuning. Along these lines, the executive can assess which artist or collection would co-ordinate the client's inclinations towards the customers. For the music lovers, music is lifeline and music are a lovely part of the everyone's life because everything in this world can be related to frequency and vibrations. According to all the good things about the music and the high demand for recommendation systems in the market, we chose to do music recommendation system.

Abstract: Crime prediction is a unique approach to identify and to find pattern trends of crime. Prediction means, using analysis and learning techniques, to find predictive actions of a specific activity and this is found to be effective in doing predictive analysis for various tasks such as crime prediction. The aim of this paper is to implement an approach for the problem in predicting the number of cases of crime happening in different parts of India. During the research we considered the machine learning model Random Forest and used the same for the prediction for crime. The prediction metrics used in this model are taken from feature selection technique. This technique increases the efficiency and accuracy of the prediction and also to avoid the model from over fitting. This model was tested on the crime data of India.

Abstract: For years, humanity has been progressing with the cost of harming the environment. And now one of the biggest change and solution being the introduction of electric vehicles. And the past few years’ electric vehicles had shown us it’s environmental and economic advantages, but distribution of the charging stations of these electric vehicles is crucial so that it could meet the needs of the users of these electric vehicles. Numerous attempts have been made to tackle this problem to find an optimize way to allocate the charging stations, but the traditional mathematical equation used are time consuming and suffers when put in new conditions such as different countries as the constants taken changes according to the places. But having the advantage of manipulating large data with the help of machine learning and applying data algorithms which adapts with different situations and bringing out hidden inferential we could take a new way of handling this problem. This paper consists of an exploration of computational ways, using machine learning algorithms to determine an optimal allocation of the electric vehicle’s charging stations in metropolitan cities and creating an interface for ease of use, also a thorough comparison with petrol pumps.

Abstract: As an agricultural country, India's economy is heavily reliant on agricultural yield growth and agroindustry goods. Food demand is rising as the world's population grows by the day. Climatic conditions are the foundation for growing the best produce. The internet technology is advancing on a regular basis, and companies are becoming digitized. Every company has a website or a mobile application that they use to give services to its customers.

Abstract: The greatest threat to humanity is water pollution. It causes affliction to animals, plants, etc. To avoid the problem in the transportation sector, we need to foretell water standards from pollution using machine learning methods. monitoring and forecasting the value of water has become a vital area to research. The goal is to examine machine learning methods for water quality forecasting by predicting the results to the best accuracy. Dataset is analysed by Supervised Machine Learning Technique (SMLT) to bag a number of details such as, variable identification: uni-variate analysis, bi-variate analysis and multivariate analysis, lost quantity treatment and data validation analysis, data purification / preparation and data detection will be performed on the dataset. The analysis provides a clean guide to examine the sharpness of the model parameters in relation to fulfilment in predicting water standard by calculating its efficiency. To offer a method to accurately predict the Water Quality Index (WQI) value by predicting the results in the form of accuracy from comparing supervised classification machine learning algorithms. Furthermore, to correlate and canvas the effectiveness of various algorithms from the given dataset with evaluation of classification report, confusion matrix, categorizing data from priority and the result shows that the performance of the suggested algorithms that can be compared with Accuracy which is done by evaluating Precision, Recall and F1 Score of the algorithm.

Abstract: Clinical decision-making in health care is even now inspired by data-driven computer forecasts or suggestions. A range of machine learning functions has recently been shown in clinical works, particularly for result prediction patterns spanning from humanity to stroke. We investigate the state of the art in relevant subjects such as data point treatment, interpretation, and simulation assessment in the framework of outcome prediction models improved utilizing data as automated health data. We also look at the flaws in widely used modeling assumptions and offer suggestions for further research

Abstract: A software bug is some sort of a fault in the source code or a computer program. These bugs work in unusual and unintended ways which is a serious problem for a programmer and the company. Detecting bugs in software has been tried and tested through multiple means, the most recent of which is Machine Learning algorithms. Using a revolutionary dataset consisting of real software code snippets of the C language into key values, we were able to train the algorithms based on numerical parameters. This in turn simplified our algorithmic process. Furthermore, we run multiple classification algorithms to gain precision, recall and Fn scores, and improve upon these scores using key hyperparameter tuning techniques. Our observations revealed an increase in accuracy and were able to create an end module which can directly take the source code as an input from which the metrics and features are extracted and give the output if the code has a software bug or not.