Understanding Farmers’ Multi-channel Communication Preference in West Bengal Using Association Rule Mining

Abstract

The present study was conducted during October, 2025 at Bandh Nabagram village, Sriniketan block, Birbhum district, West Bengal, India to investigate communication behaviours of fifty farmers using Association Rule Mining (ARM). ARM is a data mining technique to identify probabilistic relationships among channels such as Krishak Bandhu, social media, Television, Radio, Agricultural Officers/KVK, and Local Input Dealers. Field-level agricultural communication was inherently complex, as farmers rely on multiple channels whose usage patterns and interdependencies vary across individuals. Each farmer’s channel usage was encoded as a transaction, allowing ARM to generate association rules, with strength and relevance quantified through support, confidence, lift, leverage, and conviction. The analysis revealed sixty-two combinations of communication channel usage among respondent farmers. Support values ranged from 0.46 to 0.54, indicating frequent co-usage of certain channels. Confidence values spanned 0.46 to 1, highlighting deterministic and complementary pairings. Lift (1–2.17) and leverage (0–0.248) identified synergistic combinations exceeding random expectations, while conviction (1–∞) highlighted highly reliable predictive relationships. Notably, combinations such as Radio with Agricultural Officers/KVK and social media with Krishak Bandhu demonstrated perfect predictability, emphasizing their importance in extension design. The findings underscored farmers’ multi-channel information-seeking behaviour and indicated that integrated digital and interpersonal strategies optimized knowledge dissemination. Policymakers and extension agencies prioritized impactful channel combinations, enhanced adoption of innovative practices, and designed context-specific interventions in rural West Bengal. Future research was suggested to examine temporal dynamics, larger populations, and emerging digital tools.