The EvolveDTree System
Gustavo Alexandre
Niterói, State of Rio de Janeiro
- 0 Collaborators
This project proposes a system of data mining based on machine learning and genetic algorithms, in order to be able to classify the tendency of a student to abandon or graduate in the course in which he is enrolled, in a probabilistic way. ...learn more
Project status: Under Development
Intel Technologies
Intel CPU
Overview / Usage
Brazilian society suffers constant financial damage when students of higher education courses disassociate with universities without completing the courses in which they were enrolled, mainly in which there was a contribution of public resources. The academic management of an IES comprises several activities. Within the Brazilian federal universities, there are several socioeconomic policies and programs with the objective of assisting and providing support for actions that seek to minimize the dropout of enrolled students, maximize the number of students graduated in adequate course time, as well as improvements in the process of learning. One of the main challenges of higher education is a dropout, a situation that occurs when students fill the vacancies and dissociate themselves from universities without completing the course in which they have enrolled. According to OECD (2013), the average annual cost per undergraduate student in Brazilian public education was US$ 13,539.90. This value shows that evasion implies significant financial pain for the country.
Methodology / Approach
-
Exploratory Data Analysis
-
Preprocessing and Data Quality
-
Feature Engineering and Feature Selection
-
Training Models
-
Evaluation Results
-
The Prediction Model
Technologies Used
- Python
- Pandas
- Numpy
- Scikit-Learn
- DEAP
- TensorFlow
- GPU Tesla
Repository
https://github.com/gassantos/evolvedtree