Estimation of housing sales in Turkey using data mining methods

In recent years, the rapid increase in housing investments has become one of the important growth factors in the Turkish economy. This situation gave rise to the idea that forecasting housing sales could contribute to the economy by accelerating the decision-making process and facilitating correct investment planning. The increasing amount of data makes manual analysis difficult to perform. Therefore, the use of data mining applications has increased in the process of extracting information from large data sets. It is used in various fields such as data mining, prediction, classification, debugging and evaluation. This study focuses on making predictions by applying data mining techniques to housing sales data. In order to create a prediction model, 120 months of housing sales data for the periods 2013-2023 were processed and a data set was created. Relevant data, T.R. Obtained from the Central Bank's Electronic Data Distribution System (EVDS) and the Turkish Statistical Institute (TUIK). After the monthly housing sales data were prepared and cleaned for data mining in Microsoft Office Excel, modeling work was carried out in RapidMiner 10.0 software. Relative Error (RE) value was used as the performance criterion. Looking at the model performances, it was seen that the Support Vector Machine (SVM) algorithm gave the best prediction result with a RE value of 0.2%. It was also found that the Generalized Linear Model algorithm performed very well with a RE value of 0.3%. This study has the lowest error value compared to similar studies in the literature. Thus, it is thought that it will contribute to the literature and management of construction projects. Keywords: Housing Sales Numbers in Turkey, Artificial Intelligence, Data Visualization, Data Mining, Forecasting, RapidMiner