Modeling of predictive data mining algorithms in the "R" programming language and comparison of their performances

Today's economy is dynamic. With the developing information technologies, the number of recorded data has also increased. The increasing amount of data makes it difficult to see the relationships among the data. Obtaining information from raw data and using the obtained information in future predictions are of critical importance for businesses. Data estimation is an imprecise and complex process. However, estimation which is the closest to right is very important for businesses to make strategic decision. Data forecasting is widely used in economics. In this study, export data, which is of great importance for economic development, was examined. A data warehouse was created using statistics from the Turkish Statistical Institute and The Central Bank of the Republic of Turkey. Algorithms were developed using the R programming language, which is frequently preferred in statistical analysis. Artificial neural network, regression and time series algorithms were developed in R programming language. In the first stage of the study, an artificial neural network algorithm was developed in the R program. At this stage, different network topologies were tested and the most successful artificial neural network was determined. Accordingly, it was detected that the network with the (5,3) topology had the most successful performance. In the second stage of the study, a regression algorithm was developed in the R program. In the last stage of the study, the time series algorithm was developed in the R program. Naive Bayes and ARIMA models were tested and it was detected that the ARIMA(1,0,0)(2,0,0) model was more successful. Artificial neural network (5,3), regression and ARIMA(1,0,0)(2,0,0) algorithms were tested for the training data in the data warehouse. The success of the algorithms was compared by calculating the statistical error rates. Accordingly, it was concluded that the most successful prediction algorithm was the artificial neural network.