Prediction of the Numbers of Visitors at the Sinop Museums by Artificial Neural Networks lindeki Müzelere Gelen

In this study, the numbers of museums ‘visitors (Archaeology, Ethnography and Historical Prison) at the city center of Sinop province have been predicted by Artificial Neural Network structures. Artificial Neural Network models have been created in MATLAB environment. These Artificial Neural Network models are feed forward and trained by Backpropagation Algorithm. For each museum, a Artificial Neural Network with 19-inputs and 1-output have been used separately. As inputs of networks, 10 different meteorological factors, time factor (month, year), tourism income (TL), exchange rate ($/TL) and monthly-yearly PPI and CPI data have been used. Output of ANNs is the daily average of number of visitors for each month. In order to train and test the Artificial Neural Networks, the number of visitors of museum at city center for total 60 months of years between 2012 and 2017, and other input data have been used. The selection of proper Artificial Neural Networks structure have been achieved by trying backpropagation training functions 50-times on 3-different activation functions structure with 8 different neuron counts at one hidden layer. Totally, 32400-network have been created by training and the best network structure for each museum have been selected. Estimation result obtained by the Artificial Neural Network models have been evaluated and discussed. As a result of this work, it has been proved that estimation of number of visitors visiting museums at Sinop province can be done by using ANN structures.


Introduction
Estimation of tourism demand is very important for tourism and service sectors. This can provide effective information for tourism planning and policies. In order to estimate tourism demand correctly, it is essential to use a valid method. It is important to develop correct estimation methods for the continuation of research and planning of the future. Thus, with the investments to be made in the tourism sector, it will be made an important contribution to increase the employment and trade.
There are many national and international studies in the literature made on tourism demand and its estimation. In these studies, Artificial Neural Network (ANN) models have been formed by using the relation between input and output data and it has been found that ANNs provides solutions with acceptable errors. Andrawis et al., (2011)  For this problem, ANN with radial based functions was considered to be the best (Cuhadar et al.2014).
In the work of Karahan carried in 2015, monthly tourism demand of future periods has been estimated by input variables such as weather conditions, revenue from tourism, exchange data, and Consumer Price Index (CPI). According to the correlation between the tourism demand estimation produced by the suggested ANN model and the real tourism demand of the same period, it was seen that the suggested model was highly accurate. In (Ali and Shabri, 2017), the number of tourists going to Malaysia from Singapore in the years 2010-2014 are estimated by the techniques of ANN and support vector machines. Hence, they stated that ANN give better estimation than the support vector machines method.
In this work, the numbers of visitors of the museums (Archaeology, Ethnography and Historical Prison) in the city center of Sinop province have been estimated using artificial neural network (ANN) structures. In order to train and test the ANNs, the number of visitors of museum at city center for total 60 months of years between 2012 and 2017, and other input data have been used. Output of ANNs is the daily average of number of visitors for each month. Estimation results obtained by the ANN models have been evaluated and discussed. As a result of this work, it has been proved that estimation of number of visitors visiting museums at Sinop province can be done by using ANN structures successfully.

Artificial Neural Networks (ANN)
ANN is a simulated system using the human brain's ability to perform a function. ANN is composed of interconnected artificial nerve cells and is mostly in the form of layers. It is done with software on computers. It is similar to the information-processing feature of the brain. Analytical methods can solve problems that are difficult to solve effectively. ANN, can effectively solve problems that are difficult to solve by analytical methods. Many functional processes such as the rapid identification, perception, interrelation, evaluation of data of a very different structure can make an active and quickly. In this respect has been widely used in many different disciplines in recent years.
This signal is usually the output cell of another nerve cell. Each X vector is multiplied by the associated weighting factor w. As a result, the weighted vector X is obtained. All inputs and weighted X vectors come to the collection module and their algebraic totals are made; as a result of X vector, S (total function) level of influence is determined. S is obtained as the equation 1. (1) A nerve cell output signal is calculated by f (activation function). The system output of is obtained with Equation 2. (2) The General model of Artificial Neural Network is shown figure 1.

Training and Testing of Artificial Neural Network
The literature review shows that meteorological, seasonal and economic factors have effects over the number of visitors while estimating of monthly tourism demand. The number of visitors in this study have been taken from Sinop Provincial Directorate of Culture and Tourism. The variables used as input data have been given in Table 1.As the output of the network, the average numbers of daily visitors that are obtained by dividing the number of visitors per month to number of days of the same month for subjected museum have been used. In the training and testing of the created ANN models, the data of 60 months from 2012 to 2017 have been used. A total of 32400 trials have been conducted to determine the networks to be selected for 3 museums.
The best, the worst and the average R2 (determination coefficient) values in the each 50 tests of the combination for training and activation functions have been computed. Among the results, the best value for the R2 has obtained. From the resulting network set, the R2 value has been chosen to be the best. Equation 3 used in measuring network performance has been given. (3)

Findings and Discussion
The combination of the training and activation functions have been run with 50 repetitions as 5,10,15,20,25,30,35,40  function (purelin). These: Levenberg -Marquardt backpropagation, scaled conjugate gradient backpropagation and resilient back propagation and hyperbolic tangent activation function (tansig), logarithmic sigmoid activation function (logsig) are the linear activation function (purelin). In the Table 3, the best results from the combinations made for 3 different museums and their characteristics have been given.

Conclusions and Recommendations
The number of the visitors at the Sinop museums (Archaeology, Ethnography and Historical Prison) is tried to be predicted by using ANN structures in the concept of this study. . ANN models have been created in MATLAB environment. These ANN models are feedforward and trained by backpropagation algorithm. For each museum, an ANN with 19-inputs and 1-output have been used separately. As inputs of networks, 10 different meteorological factors, time factor (month, year), tourism income (TL), exchange rate ($/TL) and monthly-yearly PPI and CPI data have been used.
Output of ANNs is the daily average of number of visitors for each month. In order to train and test the ANNs, the number of visitors of museums at city center for total 60 months of years between 2012 and 2017, and other input data have been used. A total of 32400 trials have been conducted to determine the networks to be selected for 3 museums.
The best, the worst and the average R 2 (determination coefficient) values in the each 50 tests of the combination for training and activation functions have been computed.
As a result of the study, ANN method has a close estimation rather than other traditional methods and produced solutions in small error rates have been seen. ANN can be evaluated as an alternative method to traditional methods in order to predict the number of the museum visitors. ANN is also considered as a beneficial tool in establishing more fruitful business plan with accurate estimation results. . Thus, it could be possible the more the number of visitors of historical places and the richer the culture gets For future studies, it is thought that estimates that are more accurate can be obtained by creating a data set with data from a wider period.