Recently, graduation rates have been increasing while dropout rates have remained stagnant. This project looks to lower these dropout rates with the help of machine learning to predict the number of students at risk of dropping out.  For this project, student enrollment and dropout data for schools across the California and the median  income of families by ethnicity for each county in the state were first collected and then preprocessed  to extract the desired input variables for the machine learning algorithm.  Once data had been curated, feature engineering was performed and the data was then classified into different datasets for training, testing and validation of the model. The model was initially developed using linear regression, but the result accuracy was not high enough to fall into acceptable range. Instead an alternative algorithm, XGBoost (Extreme Gradient Boost) linear regression, was used and it provided better accuracy for its predictions. The output provides a  prediction value of students who are likely to dropout for each ethnicity and gender among the total number of students enrolled in the school. This prototype’s ability to predict these results can be very useful for school administrators as they can use these numbers to see which students are at risk for dropping out. It can lower the number of students who dropout since school administrators can intervene earlier and prevent them from dropping out. Additionally, the same algorithm can be extended to predict the likelihood of individual student dropout, given access to that dataset.