A modified grey-Markov fire accident model based on information turbulence indices and restricted residuals. Issue 4 (1st October 2016)

• Record Type: Journal Article
• Authors: Edem, I.E.; Oke, S.A.; Adebiyi, K.A.
• Is Part Of: International journal of management science and engineering management. Volume 11:Issue 4(2016)
• Page Start: 231
• Page End: 242
• Publication Date: 2016-10-01
• Subjects: Fire accidents -- forecasting -- fluctuation, grey-Markov -- restricted residuals -- Markov transitions; C22 -- C53 -- C60 -- J28 -- O21; Management science -- Periodicals; Engineering -- Management -- Periodicals; Engineering -- Management; Management science; Periodicals; 658.005
• DOI: 10.1080/17509653.2015.1062736
• Languages: English
• ISSNs: 1750-9661

Abstract

Abstract: This paper developed a forecasting correction model, the Modified grey-Markov model (MMSGM), for manufacturing and processing fire accident forecasts based on Mao and Sun's 2011 grey-Markov model (MSGM). The existence of a restricted relative residual state width of equal magnitude having an infinite number of restricted relative residual bounds from which Markov-chain states are created was established. The Markov cyclic chain state (MCCS) transition technique and a capture mechanism were developed and employed to capture MSGM forecasts obtained from the state creation approach. Comparative results based on out-of-sample mean absolute percentage error (MAPE) evaluation showed the superiority of the model in more than 85 and 70% of 1080 simulated data runs employed relative to the grey and grey-Markov models, respectively. MAPE difference with standard deviations of 25.26 and 11.77 (10.42 and 7.02) were obtained for the grey (grey-Markov) models, respectively, indicating superior performance. For inferior performance, MAPE difference with standard deviations of 1.82 and 3.65 (3.09 and 3.77) for the grey (grey-Markov) models, respectively, were also observed. Fire accident data application showed that MMSGM performed well with simulated and real-time outcomes, establishing that the approach is effective and suitable for fire accident forecasting in the presence of limited historical data.