A series of coal Hardgrove grindability index (HGI) tests were performed to determine each iteration's content of the monomacerite microlithotypes vitrite (Vt) and inertite (In), the bimacerite microlithotypes darite (Cl) and durite (Du), and the trimacerite microlithotypes duroclarite (Dc) and clarodurite (Cd). In HGI tests, larger particles are broken in grinding, the resulting daughter particles can be composed of different microlithotypes than the parent particle. Therefore, predicting the overall composition of the later iterations h complicated. A time series of the compositions values was constructed by sequencing the contents of all iteration (in different mesh sizes) by the order of iterations. Seasonal autoregressive integrated moving; average (SARIMA), a widelyused method of time series analysis, was employed for forecasting the final iteration's content of Vt, In, Cl, Du, Dc, and Cd. The proposed methodology was able to forecast the last iteration's composition, comparable to the actual observation. (C) 2016 Elsevier B.V. All rights reserved.