KOBİ'lere yönelik enerji verimliliği destek program başvurularının başarısının veri madenciliği ile tahmini

99.8% of the businesses operating in Turkey are small and medium business enterprises. Small and medium-sized enterprises (SMEs) also provide 76.7% of total employment. As a result, they are crucial to the economy of the nation. As a result of receiving such large payments in the economy, their energy consumption is also high. Particular attention should be paid to the energy distribution of SMEs among the employment for the limits of energy consumption. As a result, it benefits the nation's economy, the wise use of resources, and environmental consciousness. KOSGEB is an institution operating under the Ministry of Industry and Technology in order to meet the social and economic needs of Turkey, to increase the superiority and level of SMEs above the share and performance of SMEs, and to achieve industrial control suitable for economic development. In this study, it is aimed to estimate the approval status of the projects by analyzing the success and failure of the projects of the enterprises that applied within the scope of the SME Development Support Program 'Increasing Energy Efficiency' project call published by KOSGEB in 2017. Thus, the scientific approach will be taken as a basis in the decisions made about the projects and the estimations. Projects evaluated by KOSGEB in 2017 to be used in the study; company status, number of partners, number of years of operation, number of existing personnel, number of white-collar personnel, number of blue-collar personnel, project budget, the part of the project to be met with equity, the enterprise's equity, domestic sales, foreign sales, administrative expenses. The role of these qualifications on the approval of the project by KOSGEB will be examined. The project application data from 81 provinces of Turkey is used in the study. Applications are received online through the system and stored by KOSGEB. Applications are evaluated by the application evaluation board. With this review, it has been tried to make an estimation of the success of the enterprises that will apply for the projects in the future. The data were preprocessed and adapted to the python software. Data set; LightGBM, Logistic Regression and Support Vector Machines algorithms are classified. Classification results were compared and LightGBM algorithm was found to be the most successful algorithm with 0.93 F1 score and 0.88 accuracy.