- Build a smart Predictive tool that can assess the chances of a Student's entry into multiple Tier US Tech Schools.
- Develop a prototype model that uses Student data to produce admission probability for each student.
- Explore historical student data, using statistical analysis and Machine Learning, to understand and identify factors that may affect a Student's success at getting into their dream University.
After observing many of my friends trying hard and not getting into the University of their choice in the US, I was intrigued by the admissions, which from outside looks like a black-box, could actually be broken down using Analytics and be predicted by Machine Learning. When a candidate gets a low GRE or TOEFL score, it causes much anxiety and tension. So, I look whether they play as much an important role, or there are other factors that can quantitatively outweigh Standardized Test Scores. Furthermore, this tool can help students in building a stronger Profile.
-
Importing Libraries
-
Data Pre-Processing
- 2.1 Reading Data from the csv file.
- 2.2 Dummification of categorical Variables
-
Data Modeling
- 3.1 Statistical Learning
- Building a statistical Regression Model to find statistically significant factors.
- 3.1.1 Interpret the p-values (Set 5% as level of significance)
- Analyzing important Factors with significance lower than 0.05.
- 3.2 Machine Learning
- Making a Machine Learning model that will finally yield the prediction.
- 3.1 Statistical Learning
| Model | Percentage RMSE | R square | Adjusted R-square |
|---|---|---|---|
Linear Regression |
9.84 | 74.3 | 72 |
Using a very simple Linear Regression model the prediction of getting into a University is quite good. It is seen that TOEFL score is quite insignificant factor in deciding the admit of a student to any University. Whereas, GRE scores, University Ranking, Letter of Recommendation and Previous Research can play an important role in predicting the admit.