Modeling Texture Properties of Different Chocolate Varieties Using Artificial Neural Networks: Effect of Composition

ABSTRACT Texture and rheological properties are key properties that play a significant role in consumer acceptance and final product quality of chocolate and should be monitored for product quality control and appropriate process selection. This study proposes a rapid, non‐destructive alternative for determining these properties. Artificial neural network (ANN) modeling was applied to predict hardness (output) and apparent viscosity (output) based on chocolate's composition (input). Two models were developed for this purpose: ANN‐1 to predict hardness and ANN‐2 to predict apparent viscosity. The ANN models were able to predict the properties with good accuracy (ANN‐1: R 2 = 0.95, ANN‐2: R 2 = 0.97) and low mean error (MSE). This study demonstrated its effectiveness as a tool for predicting chocolate properties based on its composition. The resulting neural network model can help chocolate manufacturers predict chocolate's texture and determine its intended use in new product development, thereby improving productivity or product consistency.