Predicting the retention of customers of sport gym using the K-nearest neighbor algorithm

Today, strategic and commercial policies are focused on maintaining and improving customer retention and increasing trust in the organization. The most important reasons for such a change is the increase in public awareness and information about the consequences related to customer satisfaction and retention. The customer is considered a key and central factor in strengthening, profitability and survival of the organization, and the orientation of all the goals, strategies and resources of the organization is centered around customer attraction and retention. Hu et al. (2009) showed that retaining customers has a strong effect on the company's profit compared to attracting new customers. The expansion of sports requires attention and organization based on sports science through attracting and satisfying participants in physical activity. One of the most important small institutions Sports clubs play a fundamental and important role in the development of sports and physical activity in society. It is necessary to fully understand the components of customer satisfaction in health and fitness centers in order to effectively provide sports services to improve their health. It is necessary to identify important and effective factors in order to attract people to sports fields and to examine important factors in the continuation of their company (customer retention) in these fields by using different methods of research and research and finding the special needs and interests of people in the society Data mining is one of the most suitable options to help extract knowledge from a large volume of data, discover patterns and generate rules for predicting and comparing data that can help organizations make decisions and achieve a higher degree of confidence Data mining algorithms and considering several influential factors have been able to achieve reliable accuracy and accuracy in forecasting. For example, Keimasiet al. (2016) used C5.0 and Interactive CHAID algorithms and the information of 10,300 members of Mellat Bank customer club to cluster customers into loyal and non-loyal groups. This data includes two main parts of demographic information and information was related to the services used. Bay et al. (2022) investigated the effect of social responsibility on customer retention of fitness clubs with the mediating role of club reputation and confirmed the effect of social responsibility on customer retention with the mediating role of club reputation and reported that performing social responsibilities by gyms can play an important role in the reputation of the gym and increase their retention and manually selected 10 factors (traits) with the greatest impact on results.Davoodi and Khanteymoori used an artificial neural network with five learning algorithms to predict the results of horse racing. They found that the conjugate gradient method is the most suitable algorithm for predicting the last horse. Because of the growing awareness of health and fitness, the gym industry is booming. As a result, competition between private club businesses is becoming increasingly fierce. Therefore, it is critical for private clubs to develop membership retention strategies to prevent their customers from switching to a competitor. Using technology to predict churn can be beneficial for private clubs looking to stay on top of the business. According to evidence, no study has used data mining algorithms to predict customer satisfaction and retention of private clubs. Therefore, the purpose of this article was to predict the loss and retention of customers of private clubs using the nearest neighbor algorithm.
The statistical population of this study was related to 724 athletes who participated in the online call (WhatsApp, Instagram, Telegram, etc.) by completing the questionnaire in the present study. Of these, 640 athletes answered the questionnaire, of which 103 (16%) were confused. Finally, the sample included 537 athletes, 257 (48%) were women and 280 (52%) were men. This study was completed using an anonymous, researcher-made electronic questionnaire that had 18 factors related to customer satisfaction, and its validity was checked by 5 university professors and experts in the related field. In this data mining, k-nearest neighbor (KNN) algorithm was used for prediction. In this method, the data were first divided into two groups: training (for training) and test (for evaluation). The classification of data was done completely randomly using the partition node in the training data of Clementine software, where the training data had 75% of the data and the test data had 25% of the rest of the data.
The findings indicate that the support vector machine algorithm can predict the percentage of permanent and non-permanent people with 73.4% accuracy and 71.6% accuracy using 18 customer retention factors. In this research, for the first time, the model of predicting customer retention in the field of sports venues has been investigated and, unlike other researches, the factors that may affect customer churn and retention have been analyzed and collected. The current observation suggests that the knowledge of the trainer, the up-to-dateness of the equipment, the availability of the hall and the use of social networks, etc. can be used to evaluate the tendencies of the client to drop. Also, as an industry whose sustainability is dependent on customer retention, they should place great importance on existing loyal customers, while at the same time they can introduce exclusive resources such as a full-time knowledgeable trainer to attract random customers. In addition, analyzing customer preferences and providing better and more convenient customer service should also be used as a marketing strategy. Prediction accuracy can be improved by using machine learning and data mining techniques that have not been used in this field, but have shown good results in other fields. Also, the use of combined algorithms can increase the prediction accuracy. In addition, the use of various factors and features such as the price performance of club membership helps to make more accurate predictions. On the other hand, with the help of a group of specialists and managers from other provinces of the country, a comprehensive data set can be collected in each sport. In order to provide an opportunity for comparison between different studies, researchers are advised to collect data from successful private clubs with high membership.