چكيده به لاتين
Retirement season always is considered a momentous period of people’s life that has various individual and social aspects. Changes in the living conditions of retired people in comparison to their employment age, bring in new issues to their lives. Different organizations and individuals have different views on this issue, because it could have different effects to them. After retirement longevity and factors that could likely affect it are two remarkable aspects in this context. In this study, an artificial neural network (ANN) model is developed to predict the longevity of retirees, based on their occupational, health, and personal factors. The model was based on data from 351 retirees of an Iranian steel firm who retired between 1371 and 1392 (Iraninan calendar). In order to ensure the accuracy and performance of the model, its forecasting result is compared to a multiple regression model (MR) in terms of their forecasting accuracy by using a relative mesure known as mean square error (MSE).The forecasting error of ANN model is found to be 4.8 % of that derived from MR model. In the second part of the study, the importance of factors affecting longevity were identified using simultaneous regression.Retirment age and pre-retirement income are found to be strong predictors , whereas being smoker, being diabetic , and having previous surgery on vital organs are considered as weak predictors of longevity. By eliminating the age of retirement from the study, number of children, level of job hardness, pre-retirement income, and being smoker had moderate impacts on retirees’ longevity.
