Stock Prediction & Recommendation System Using KNN And Linear Regression

Abstract

Machine learning is a method of data analysis which tries to find a pattern in a data set by analysing the features of the various data points present. Since, there are so many datasets available, learning data analysis will become a necessary condition for the advancement of the technology. People who invest their money have to come across the hurdle of predicting day-to-day prices, which helps them invest in the stock market honestly by taking risks and variations in assumptions. In this project, we will use the KNN method and direct retrospective analysis of stocks and predict their closing value based on previous year's performance. Linear Regression performs better than KNN on selected dataset. On the basis of the model's outcomes, people can buy/sell stocks accordingly. The report sets out the process for analysing and estimating stock prices over the past year. This approach consists of three steps, namely, data collection and integration, data classification, analysis and prediction. After data classification, linear regression is used to analyse trends within each collection. The results from the model are analysed along with the other practical trading methods. Despite the volatile nature, the efforts of using machine learning for analysis and prediction showed some success. For much more enhanced prediction system, further analysis of results could help. The proposed method is not only limited to stock trading but can also be applied for trend analysis in any time series data.