نوع مقاله : مقاله پژوهشی
نویسندگان
1 دانشجوی دکتری مدیریت ورزشی، دانشکده علوم ورزشی، دانشگاه اصفهان، اصفهان، ایران
2 دانشیار مدیریت ورزشی، دانشکده علوم ورزشی، دانشگاه اصفهان، اصفهان، ایران
3 استادیار مدیریت ورزشی، دانشکده علوم ورزشی، دانشگاه اصفهان، اصفهان، ایران
4 استادیار بیومکانیک ورزشی، دانشکده علوم ورزشی، دانشگاه اصفهان، اصفهان، ایران
کلیدواژهها
موضوعات
عنوان مقاله English
نویسندگان English
The purpose of this study was to estimate the prices of Iranian professional soccer league players, as well as to compare the regression methods and neural networks in predict that. The research method was mixed exploratory designs, which is a combination of qualitative and quantitative methods. The statistical population of the research in the qualitative sector consisted of managers, coaches of clubs and experts familiar with the field of player buying and selling that fourteen people were selected by snowball method until reaching the saturation point. The statistical population in the quantitative section also included all footballers present in the Persian Gulf Football Professionals League in the years 2017-2018, which 226 of them were selected by Cochran formula and selected by stratified random sampling. The research instrument in qualitative method included in-depth interview with statistical sample whose reliability was calculated 81% through test-retest method. The data needed for quantitative analysis were also collected from reputable sites and the Iranian Football League. Pearson correlation coefficient, linear regression and radial neural networks were used to predict and design the model using SPSS software version 21 and R version 3.6.2. Qualitative research findings showed that player performance, personal characteristics, abilities, club characteristics and bubble-generating factors are effective in determining the price of football players. The results indicated that, despite the benefits of regression methods in predicting problems, because of the multifaceted and complex pricing of soccer players, artificial neural network models would cover more variables and be more efficient and accurate.
کلیدواژهها English