Big data analysis model of strategic agricultural food products.

Fresh agricultural products can be a commodity that determines inflation in a country. The government as a regulator should be able to monitor the prices of strategic agricultural food products. Indonesia already has a national strategic food price monitoring system that is recorded. In this research, big data analysis is integrated with knowledge discovery in the database with an open government maturity model in the initial step. Various analytical techniques used are k-means, decision tree, naïve Bayes, recurrent neural network (RNN), regression, and model explanation. The study focuses on selling prices at the consumer level for strategic horticultural commodities such as red chili, cayenne pepper, shallot, and garlic. This study provides descriptive, diagnostic, predictive, and prescriptive analysis models for strategic agricultural food products. The results of the descriptive analysis show that the k-means algorithm has a good performance in terms of price clustering compared to standard deviation. The diagnostic analysis results show that the amount of production and the price at the farmer affect the price level at the consumer. The decision tree is the best alternative algorithm for classification with an accuracy above 85%. The RNN model is the best model for predicting fluctuating prices with an accuracy rate above 96%. The results of the prescriptive analysis are presented in the form of a development plan to support sustainable farming. Integrating these methods can be used and developed for comprehensive big data analysis research in various fields. [ABSTRACT FROM AUTHOR]