Korean J Occup Environ Med.  2000 Dec;12(4):515-523.

The Prediction Model of the Number of Industrial Injured Persons Using Data Mining

• Yi KH
• Chung HK
• Park JS

Abstract

OBJECTIVES
This study is to see the transition and pattern of the industrial iureal worker, and to develop the prediction model.
METHODS
The data of the study are based on the samples from data-warehouse of Occupational Safety & Health Research Institute and are summed monthly from Jan 1986 to Dec 1999. This study data used data mart and Meta data from DW in KOSHA. The prediction model of the injured worker in Industry is designed by using a winters time series method after data preparing (i. e. sample, explore, modify) from DW.
RESULTS
Thls predicted model obtained Winters-method multiplicative in exponential smoothing among applied all models, after the tlme series (total 163 months). It showed that the prediction power was 95.5 %.
CONCLUSIONS
In the process of exploring the data, totally the rate of industrial injureal workers reduced, and in the yearly circulation, in February and September the number is the lowest but in June, July, October and November the higher. The number of monthly average injureal workers is 8709 (95 % confidence interval 8277, 9140). From the developed prediction model, since Aug 1999 the industrial injureal worker reduced rapidly in Dec 1999 and first period of 2000. But In second period of 2000 the number of the injured workers is increasing. To conclude, as the total economic situation is becoming better in 2000 than In 1999, its is supposed that the injured workers will increase more than the predictive injured workers because of the increase of production rate and labor force.

MeSH Terms

Academies and Institutes
Data Mining*
Employment
Humans
Occupational Health