Sales forecasting in public procurement sector with statistical data mining techniques

In today's economic conditions, states must cooperate with the private sector in economic situations where it is not possible for them to offer public services. At this point where states and private sector meet, the concept of public procurement is emerging. Today public procurement is carried out within the framework of "Public Procurement Law". Public procurement, which plays an important role in improving the quality of services provided by the state, indirect savings, encouraging innovation in the private sector, closing the current account deficit and reducing corruption, undoubtedly has a very important share for the country's economies. In this study, it was aimed to forecast the sales of State Supply Office, which is authorized as central procurement institution within the scope of "exception" concept and to act as intermediary between private sector and public sector, and to investigate the sensitivities of these forecast in order to sustain public procurement effectively and efficiently with various parameters. These data, forecasted by statistical data mining techniques, will be integrated into the decision support system and the strategic goals of the institution will be realized and followed. In addition, it was planned to model the sales forecasts with linear model, generalized linear model and random trees and xgboost trees algorithms which ara decision trees , using the historical data of 2007-2018 . Thus, it is considered that the key performance indicators included in the institutional strategic plan will be provided and academic contribution will be made to increase the efficiency of public procurement sector.