Development of short-term prediction with regard to a number of accidents at work using the scoring method

Abstract

The mining industry is an industry branch with one of the highest rates of accidents at work in Poland and the presented analysis develops the knowledge about the safety in the mining sector. The work below presents a short-term prediction of the overall work accident number in a selected industrial facility, developed on the basis of statistical accident rate data and using 25 selected econometric models. In the summary assessment of a specific prediction, the scoring method was applied, taking the following weights into consideration: C1 and C2 criteria (C) – 10 % each, C3 and C4 criteria – 20% each, and C5 criterion – 40 %, where: C1 was the value of ex post prediction error  for the series including the empirical data covering the period between 2007 and 2016; C2 was the value of ex post prediction error  for the series including the empirical data covering the period between 2007 and 2018; C3 was the value of coefficient of random variation Ve for the ex post predictions from the period between 2007 and 2016 (for all predictions except the linear and linearized models, the RMSE* value was applied to estimate their value); C4 was the value of coefficient of random variation Ve for the ex post predictions from the period between 2007 and 2018 (for all predictions except the linear and linearized models, the RMSE* value was applied to estimate their value); C5 was the value of ex post prediction error  for the series including the empirical data covering the period between 2017 and 2018. Statistical work accident rate data covering the period between 2007 and 2018 were used in the analysis.