<p>A workflow optimization for electrofacies prediction in carbonate reservoirs using Self-Organizing Maps (SOM) is established in the current study to enhance the accuracy and efficiency of facies classification within complex carbonate formations. The method integrates stochastic analysis with petrophysical properties to derive electrofacies based on texture, heterogeneity, and fracture identification. SOM, an unsupervised neural network algorithm, is applied to classify multivariate input data into five distinct categories by analyzing the distribution of input variables at each depth interval. This process generates classification flags and cumulative curves, which are then interpreted as electrofacies at the individual well level. Each electrofacies is assigned to a unique color and code and subsequently mapped to corresponding lithofacies. These are further annotated with detailed lithological and textural descriptions to improve geological interpretation. Curves specific to each electrofacies are created from the SOM outputs to quantitatively represent their distribution and properties. Core data are used to calibrate and validate the SOM-derived electrofacies, ensuring the reliability of the results. The identified electrofacies range from Facies 5 (Packstone, very good reservoir potential), Facies 4 (Wackestone grading to Packstone, good reservoir potential), Facies 3 (Wackestone, moderate reservoir potential), Facies 2 (Wackestone grading to Mudstone, fair reservoir potential), to Facies 1 (Mudstone, poor reservoir quality). These facies occur in a cyclic pattern throughout the logged intervals of the Lockhart Formation, reflecting the stratigraphic and depositional variability characteristic of carbonate reservoirs.</p>