Abstract:
In year 2020 the whole world faced an unexpected crisis which we known as COVID-19
pandemic. After pandemic declaration by WHO (World Health Organization), every nation
around the world started locked down their nations and their communications to other nations
to prevent the outbreak of COVID-19. Many sectors hampered by this pandemic, as well as
education sector, especially under develop country like Bangladesh are facing a huge loss in
education sector because of COVID-19 pandemic. All educational institutes started
normalizing their schedules after two years of lockdown. And this made a huge study gap to
our students of Bangladesh, especially primary and secondary level students are facing
enormous problems to continuing their studies. Primary level students forgot how to spell, how
to pronounce how to read and how to write. Some students got addicted with smartphones and
online games and some students dropped out from school for various reasons. Some of them
dropped out for the financial condition of their family, some of them dropped out because they
lost interest in study, some of them got married and some of them started working so they could
contribute to their family to get rid of poverty. Peoples who were living their life under poverty
or who had stable financial condition before the pandemic but after pandemic financially they
are facing loss or became broke have more chances to drop out from school. By being so close
of a secondary school it’s motivated me to develop a Machine learning model by using machine
learning algorithm by seeing vast amount of dropout rate of school students after COVID-19.
In this research I applied different machine learning algorithms such as, linear regression,
Decision tree, SVM (Support Vector Machine), Random Forest, Naïve Bayes. But from all of them Random Forest got the highest accuracy of 87%. The goal of this research is mining
significant facts of being dropout from school, and to predict is any students will be dropout or
not. The proposed model was built on Google Colab (python-based ide) and trained on
secondary data which was collected through students from different secondary level school.
The dataset contains 300 data collected from students with 9 attribute of student data.
