Using Additional Sets of Random Data to Assess Possible Dissaving Risk Against Life Expectancy for Elderly People

In order to detect if the economic activity of elderly people aged sixty-five or older has started to decline could be observed, anonymous data (AD) were utilized as analysis data, which were obtained from the National Survey of Family Income and Expenditure published by the Ministry of Internal Affairs and Communications. So far, a method to detect dissaving risk among elderly people has been developed by applying three methods based on income and savings. With a view to reinforcing the anonymity of AD, random data (RD) were generated based on AD. One set of RD were set as training data with a view to performance evaluation of AD. This result has indicated that using both RD and AD would be as effective as utilizing mere AD in evaluating the performance of the proposed method. With the intention to strengthen this tendency, in this paper further additional four sets of RD are used as training data to assess the performance of proposed method. As a result of the analysis, it could be concluded that either of five sets of RD based on one set of AD would be applicable in assessing the performance of the proposed method.