Absences-v-Grade-Machine-Learning

This program was made as an experiment for Machine Learning (ML), the point of the algorithm is to see if an ML algorithm can accurately predict students final grades based on information such as absences, semester 1 grade, gender, previous failures, and study time (Optional), the importance of this is that Mountain View High School made the claim that absences negatively influence grades, I would like to use a non bias ML algorithm to prove this claim while also looking at other data to avoid claims of flukes. The way the ML algorithm learns is through looking at the data without seeing the final grade and only looking at 10% of the data to insure the algorithm does not just memorize the answers. At first it makes a random guess of the grades and then checks if it was correct or not, and uses that information to see what factors influence the grades; then, using a linear regression model learns what most influences the final grade. Based on that information the ML algorithm can make guesses to what the student grade will be with an accuracy of 96%. Using what the algorithm considers important to the final grade can tell us a lot of what truly affects students grades and what is just a fluke. Note this is a very simple first generation machine learning method and data will be affected based on the 10% the data randomly selects drastically. The algorithm is programmed in Python 3.6 and is using the modules matplotlib for graphing, pandas to have the program read the student data, numpy to manage the X and Y axis arrays, sklearn to have the algorithm learn based on training data on 10% of the data, pickle to store the algorithms data from the highest accuracy result.