Developing Eligibility Predicting Model for Applicants of Oromia Micro and Small Enterprises Agency using Machine Learning Approach

Abstract

Micro and Small Enterprise is the techniques that Oromia Regional state is implementing to reduce jobless people by organizing people under into different groups depending on the eligibility of the individuals. Currently, the eligibility of individuals desired to be organized as enterprise is identified manually. For this study, Data collected from Oromia Micro and Small Enterprise office in hardcopy format. Researcher converted data into electronic text document. Since data collected is large in size, has important data, researcher implement feature Extraction and feature selection techniques to prepare data set. From feature extraction researcher observed that eight attributes Applicant.ID, Education_status, Employement_status, Jobless_Certificate, organization_name, sectors, and types are important whereas applicants.name, Efficiency, Id given_from and Interest are attributes are unimportant. Dataset prepared from attributes confirmed as important and saved by MSDS.csv to make it more supportable format. Researcher implemented machine learning algorithm such as SMO, Naïve Bayes, and Bayes Net to build model that predicts eligibility applicants for Micro and Small Enterprise. Experiment shown SMO algorithm scored high accuracy. Therefore, finally the eligibility prediction prototype for Micro and Small Enterprise of Oromia built by using SMO algorithm.