TIME SERIES ANALYSIS AND DATA MINING FOR ROBUST SHORT-TERM COMPRESSIVE STRENGTH FORECASTING IN GEOPOLYMER MORTAR: AN ARIMA MODEL DATADRIVEN APPROACH


Yılmaz Y., Çakmak T.

1st International Future Engineering Conference, Şırnak, Turkey, 25 - 26 December 2023, pp.24-30

  • Publication Type: Conference Paper / Full Text
  • City: Şırnak
  • Country: Turkey
  • Page Numbers: pp.24-30
  • Recep Tayyip Erdoğan University Affiliated: Yes

Abstract

With a primary focus on leveraging time-series analysis and data mining methodologies, this paper delves

into the realm of geopolymer mortars, an eco-friendly and energy-efficient construction material. The target

of the study is to specify the variables that affect the compressive strength of geopolymer mortars and then

utilize those variables to estimate the mechanical property (compressive strength (CS)) of mortars in the

future. The study analyzed the effects of factors such as compressive strength, obsidian, glass waste, fly ash,

and heat on the CS of mortar. ARIMA model is adeptly employed to forecast the short-term compressive

strength of geopolymers endowed with five distinct properties and subjected to varying temperatures. A 30-

day compressive strength prediction was made based on the 28 day compressive strengths. As a result of the

analysis, the highest RMSE, MAE, MAPE and R2 values were obtained as 1.207, 0.983, 0.025 and 0.941,

respectively. Those results emphasize how well-sophisticated statistical modelling approaches could be

employed to understand and estimate the dynamics of compressive strength in geopolymer mortars.