The purpose of this project is to predict Third, Fourth, and Fifth Grade Achievement Levels on the Florida Standards Assessment (FSA) in ELA (English Lanugage Arts) and Math. The FSA is a high-stakes test that is used to determine teacher ratings, school grades, Third Grade promotion requirements, and graduation eligibility. Florida teachers spend a great deal of time using data to drive their instruction. Educators provide targeted interventions that close learning gaps to ensure students achieve proficiency of grade level standards. This model will use historical diagnostic data as well as demographics, attendance, and discipline data to predict student achievement on the FSA. The analysis of this data will provide insight as to which factors impact performance on the assessment. Schools will be able to use this model to predict achievement levels and prepare their staff and students for success.
We choose the Random Forest classifier for this project. This model is best suited for predicting scores. It is also easy to understand and explain outcomes with this model. The Scikit-Learn Standard Scaler was used to scale the numerical and categorical data. The current accuracy score for the English Language Arts model is 79.7% for predicting pass/fail. The current accuracy score for the Math model is 89.1% for predicting pass/fail.
The data was analyzed and filtered by attendance in both subject areas ELA and Math. The attendance data refers to if a student has had 10 or more absences in that school year. In ELA the data shows that scores were higher amongst students with no attendance concerns. In Math it should a similar result. Students who attend school regularly will have a higher chance of passing the FSA with a 3 or higher.
A score of 1 or 2 is not considered passing, and a Third Grade student who scores a 1 on the ELA assessment may be retained and provided remedial instruction in Reading the following year. A score of 3 or above is considered passing.
