Launching A WebApp To Predict Your Chance For Graduate Admit

Powered By Machine Learning

Photo by Vasily Koloda on Unsplash

Where there is data, there can be a practical use of machine learning. The education sector is in no shortage of the same. Here we explore the possibility of predicting the likelihood that a student, given his/her profile stats like GRE Score, TOEFL Score, Research Experience, and other parameters can get a graduate admission to the University of his/her choice.

The entire application is packaged into a web app and hosted via Heroku, which the users can access and use. Models were trained on a relatively small dataset of Indian students and the best performing model is used to make the predictions in the web app on new data points.

Here are some details of the inputs the user will need to provide:

1. GRE Score (out of 340)
2. TOEFL Score (Out of 120)
3. University Rating (Out of 5). If you are not sure about the exact rating of your target university, you can approximately estimate them and give the input. (For example, all Ivy league colleges come under the rating of 5, and so on. In general, the top 10 universities in rank lists can be assumed to be a 5, 11 to 20, a 4 and so on…)
4. Statement of Purpose (Out of 5, can be decimal. Eg. 3.6). Rate your SOP out of 5 points. You can either ask your professor, friend or self-rate yourself based on how strong you feel your SOP is.
5. Letter of recommendation (Out of 5, can be decimal. Eg. 3.6). You can rate it similar to how you did it for SOP. A couple of great recommendation letters from top-notch professors could be rated a 5.
6. Undergraduate GPA (out of 10)
7. Research Experience. This will be a yes if you had previously done any research internships/projects and published any papers in scientific journals.

Here is the link to the web app: Graduate Admit Predictor

Note: The application may hang when the traffic is high. Please wait for a while and try later if you encounter the same.

Disclaimer: This prediction is only to give you an estimate of where your profile stands and a probability of your admit. Your actual real-life result may vary. The web app is only an experimental prototype to demonstrate one of the possible use cases of machine learning in the educational sector and the results obtained are not to be taken as absolute. Do write to us about the app's performance and when its prediction diverts tremendously.

With that being said, the app can be made more robust by training it on a larger dataset of students stats and use more tractable features. If any educational service based organization, counseling center or a non-profit organization is interested in collaborating to take forward this project by providing necessary resources, feel free to reach out to us via email.

Editorial Note:

This web app was built by Abhimanyu Thakre in association with The Research Nest. Feel free to write to us at the.research.nest@gmail.com for any feedback or suggestions. Thank you!