Hydraulic fracturing in carbonate reservoirs presents unique challenges due to their complex pore structures and heterogeneous mechanical properties. This paper explores the application of statistical methods to improve fracture prediction and optimization in carbonate formations. Hydraulic fracturing is actively carried out on these formations. In order to properly plan hydraulic fracturing, it is necessary to identify the main factors affecting oil production after hydraulic fracturing. This study introduces an integrated framework combining information amount theory (IAT) and Gray relational analysis (GRA) to identify and rank the dominant parameters controlling hydraulic fracturing performance in heterogeneous carbonate reservoirs. Based on a dataset of twenty-one fractured wells in the Perme region, twelve geological and operational parameters were evaluated to determine their impact on post-fracturing oil production rate. Results consistently indicate that fracturing fluid volume and fracture width exert the greatest influence, while fracture length ranks lower due to the complex fracture networks typical of carbonates. The proposed IAT-GRA method offers a computationally efficient, interpretable tool for data-limited reservoirs, and the findings provide clear engineering guidelines for optimizing hydraulic fracturing design and execution. The study used the theory of the amount of information and the gray method to identify the main factors influencing the results of hydraulic fracturing. Having data on core parameters before hydraulic fracturing, it is possible to predict the results of hydraulic fracturing with a high degree of reliability. The regression model is based on the method of multiple linear regression. Oil production after hydraulic fracturing increases with an increase in the flow rate of oil after hydraulic fracturing and the width of the crack. Hydraulic fracturing creates multiple cracks and microcracks, forming a complex network of cracks in the formation. Therefore, the use of statistical methods helps to make an operational assessment of the result, but does not negate the use of more accurate and complex models.
Poplygin et al. (Wed,) studied this question.