Prediction of Consumers' Characteristics in Multichannel Selection Based on Hidden Markov Model: Taking Malaysian Consumers as an Example

Traditional static algorithmic models often struggle to effectively capture the dynamic and evolving nature of consumer behavior in multichannel shopping environments, where consumers frequently switch between various platforms, such as online stores, physical locations, and mobile apps. This study applies Hidden Markov Models (HMM) to provide a more accurate prediction and analysis of consumer multichannel choices. By gathering and analyzing detailed shopping data from Malaysian consumers, HMM models consumer behavior transitions across different channels, estimating critical parameters like state transition probabilities and observation probabilities. The results demonstrate high predictive accuracy, with both accuracy and F1 scores ranging from 0.97 to 0.99. These findings show that HMM can offer a more nuanced and comprehensive understanding of consumer behavior in multichannel settings. This improved modeling helps businesses optimize their omnichannel strategies, allowing them to better forecast consumer actions and align their operations with evolving customer preferences in diverse shopping environments.