Analyzed GOOGL stock closing price data set by applying various regression models and comparing them on the basis of error calculations.
The data set of GOOGL's closing price is taken and is trained under Linear Regression, Decision Forest and Boosted Decision Tree regression models. The scored values are taken and a graph is plotted between the original and the scored values using Microsoft Excel. Various errors are calculated and the conclusion is drawn for the performance of the three regression models. The Input table(2).csv file which is the GOOGL stock price values, can be downloaded by opening the project in the studio. The data is split into two parts for training and testing (70:30). The output from the models run on the two split data can be visualized in Microsoft Excel by calculating the [Mean Absolute Error] and [Root Mean Square Error]. Conclusions drawn are: - Linear Regression is the base of all regression models and provides results that are near to satisfaction but with a greater deviation from original data. - Boosted Decision Tree Regression gives a distorted line which is not suitable for our case in predicting Stock Price which has to be very near to the actual solution. - Decision Forest Tree Regression model stands out to be the best model for Stock Price with the least error. This project is made with great support and help from Prof. Sanjiban Sekhar Roy, Vellore Institute of Technology, Vellore, TN, India. : https://en.wikipedia.org/wiki/Mean_absolute_error : https://en.wikipedia.org/wiki/Root-mean-square_deviation