GRADUATE ADMISSION PREDICTOR

1 0
  • 0 Collaborators

An Interactive Machine Learning Project that determines the rate of chances of admission for an individual based on pre-requisite scores like CGPA, GRE, TOEFL, LOR, SOP etc. ...learn more

Project status: Published/In Market

Artificial Intelligence

Code Samples [1]

Overview / Usage

Developed a practical Machine Learning Model that dynamically predicts the rate of chance of an individual for getting an admission at a specific university of his interest based on the features such as CGPA, GRE, SOP, LOR, University Rating etc.

Abroad Education is a dream for every under-graduation student in India. But He/She is unaware of how to maintain a good striking score for easily cracking the admission process. Since the universities in abroad precisely observe the scores that an individual secured during his/her educational career like CGPA score and the pre-requisite scores 'GRE', 'LOR','SOP' and finally the Rating of the University that the individual is seeking to get an admission for.

Based on all the factors and the University Rating that the individual is aspiring for, I designed a Machine Learning Model that accurately predicts the rate of chance of an individual getting an admission in a particular university.

It ultimately gives a broad knowledge, how the chances of an individual for getting an admission in an abroad university precisely requires more hard work in securing good scores in the under-graduation stage as well as exposing himself to the opportunities that are more possible for him.

Methodology / Approach

Precisely the entire project is possible only through Python, because Python has a wide range of libraries that support handling of data, creating inferences from the data, applying dynamic algorithms in Machine Learning by just simply importing the necessary algorithmic implementation from SCIKIT-LEARN library from Python, that takes care of all the stuff in the background.

When working with data it is very important to note that the data is clean and less noisy enough so that the algorithms can derive more patterns and show great performance. Hence Pre-Processing is an important stage in Machine Learning, which dynamically prunes out irrelevant features by the application of FEATURE SELECTION, for which I applied Recursive Feature Elimination(RFE) technique that recursively finds the potential data points that have a high correlation or simply that contribute to the high performance of the algorithm against the input data being feeded.

After the Pre-Processing phase, it is very important to set a BENCHMARK model that initially feeds on the raw data and makes inferences out of it. The performance of the BENCHMARK model is noted by the metric -> R2_SCORE

And the core part of the project, diverse Machine Learning Models are being applied to the raw data and their performances i.e., R2_SCORES are noted and eventually compared, the algorithm or the model that has better performance is noted and further optimization is performed by the application of GRIDSEARCHCV of SCIKIT-LEARN which dynamically improves the algorithmic performance and it's performance score is noted.

Finally, the BENCHMARK model and the OPTIMIZED MODEL are compared in terms of performances and the model that gives the best performance against all the odds is selected as the best model.

The best model is then tested against the unseen data and the performance is pretty much awesome that it shows the clear picture of how the things are being correlated and the result of the model's performance i.e., R2_SCORE.

Technologies Used

Artificial Intelligence
Machine Learning
Feature Selection
Cross-Validation
Python
Scikit-Learn
Exploratory Data Analysis ( EDA )
Random Forests
Boosting Algorithms - XGBOOSTING, GRADIENT BOOSTING
Recursive Feature Elimination ( RFE )
Grid Search CV
Co-Efficient of Determination -> R2_SCORE

Repository

https://goo.gl/gFrrhJ

Comments (0)