Abstract
In the tourism sector, accommodation business demand forecasting provides a great benefit for tourism professionals, especially hotel managers, in the strategic decision-making process. For demand estimation, the artificial neural networks (ANN) method, which works similar to a human brain cell and makes realistic predictions, has been preferred. The aim of this study was to develop an eight input and output variable of the feedforward radiated back an ANN is in a specially certified hotel room occupancy rate in Turkey to investigate the applicability of the method to predict. Four different alternative network structures were created from the data set with the K-fold cross validation method. As a result of the test simulation, it was determined that the estimated and actual occupancy rates of the network with the lowest error were close to each other. According to this designed model, the monthly occupancy rate for the years 2019 and 2020 has been estimated. As a result, the effect of COVID-19 was revealed by comparing the hotel occupancy rate with the actual rates.
TopIntroduction
Forecasting is the knowledge and art that predicts what will happen in the future in the present. When making predictions, there should be historical data and information. Historical data and information are behaviors that occurred in various situations at that time in the past (Wardani et al., 2019: 1). Forecasting is about predicting the future as accurately as possible, given all of the information available, including historical data and knowledge of any future events that might impact the forecasts (Hyndman and Athanasopoulos 2018: 14).
Knowing better about the future has affected many people for thousands of years. Estimation methods vary widely. The estimation methods will depend on the availability of data, the quality of existing models, and the types of assumptions made, among others. Overall, it's not an easy task to guess. Therefore, exploring this topic is very attractive for many researchers. Artificial Neural Network has an increasing importance in prediction theory. ANN can learn from examples (historical data). ANN can recognize a hidden pattern in historical observations and use them to predict future values. In addition, it can cope with missing information or noisy data. It can be very effective, especially when it is not possible to define the rules or steps that lead to the solution of a problem (Shamsuddin et al., 2008: 1-2).
Forecasts serve a variety of purposes in the hotel industry; It provides vital inputs for marketing and pricing strategies, routine budget planning and capital investment decisions. In particular, demand forecasts play an important role in revenue management strategies commonly used in the hospitality industry. Accurate estimation of future demand is particularly important for hotels as services cannot be stocked (Walke and Fullerton Jr, 2019: 179-180). Occupancy rate estimation in the accommodation sector is of central importance in planning and decision making. Because anticipating the demand allows managers to plan on issues such as inventory, workforce, materials, financial budgeting and pricing. All of these are used to maximize revenue and minimize costs (Caicedo-Torres and Payares, 2016: 201). An accurate occupancy rate estimation will assist hotel managers, especially managers in budget-constrained hotels, in their strategic, tactical and operational planning. Accurate estimation of room occupancy rates will facilitate strategic planning and improve decision-making processes of hotel management companies (Law, 1998: 234).
When the studies on tourism demand forecasting in our country are examined, it is seen that the tourism demand of countries or cities is mostly focused. In fact, the accurate demand forecast of the accommodation businesses will ensure that the demand forecasts of countries or regions are so consistent. Hospitality businesses are the primary source of income in the tourism industry (Ulucan and Kızılırmak, 2018: 90).
There are studies of Çuhadar and Kayacan (2005) and Ulucan and Kızılırmak (2018) regarding the occupancy rate of accommodation establishments in our country.
When the studies on the occupancy rate of accommodation establishments in foreign literature are examined; Law (1998), Du et al. (2007), Caicedo-Torres and Payares (2016), Desa and Marzuki (2019), Walke and Fullerton Jr (2019).
Key Terms in this Chapter
Artificial Neural Networks: Are computational networks that try to simulate the decision process in the nerve cell (neurons) networks of the biological (human or animal) central nervous system. This simulation is a cell-to-cell (neuron-neuron, element-to-element) simulation. Inspired by the neurophysiological knowledge of biological neurons and networks of such biological neurons ( Graupe, 2013 ).
Mean Absolute Percentage Error: Is a measure of the accuracy of the predicted results ( Sood and Jain, 2017 ). In the literature, those with a MAPE value of less than 10% are “very good”, those between 10% and 20% are “good”, those between 20% and 50% are “acceptable” and those with a MAPE value above 50% evaluated as “false or erroneous” ( Ulucan & Kizilirmak, 2018 ).
Occupancy Rate: Is a measures of the success of an accommodation structure (Albu & Pacurar, 2019).
Forecasting: Is about predicting the future as accurately as possible, given all of the information available, including historical data and knowledge of any future events that might impact the forecasts ( Hyndman & Athanasopoulos, 2018 ).
Feed Forward Back Propagation Algorithm: Consists of two stages. These are a forward feed step and a backward advance step in which changes are made to the coupling strengths based on the differences between the calculated and observed information signals in the output unit ( Kizilaslan et al., 2014 ).
Tourism Demand: Is a measure of visitors’ use of a good or service ( Frechtling, 2012 ).
Artificial Neural Cell: Is the smallest basic unit in the Artificial Neural Network system (Yavuz & Deveci, 2012 AU14: The in-text citation "Yavuz & Deveci, 2012" is not in the reference list. Please correct the citation, add the reference to the list, or delete the citation. ).
Mean Squared Error: Is a measure of the accuracy of the predicted results ( Sood and Jain, 2017 ). If the MSE is zero, it means there is no error ( Al-Shayea, 2011 ).