Food trend based on social media for big data analysis using K-mean clustering and SAW: A case study on yogyakarta culinary industry

Tracking customer preferences is an important aspect of business success. Having information on hand about most favorite food is a key success for everyone who takes apart in the culinary business. Exact sales data on certain food is hardly available to the public. Restaurant owner tends to keep their data for their own business strategy. Therefore, generating a food trend in a certain community is hardly possible using food sales data. This paper discussed extracting food general trend from social media, with the case study on Twitter data with a certain regional area of interest. Social media provides a tremendous amount of data including people choice of food when they visit the certain place. However, the available data is unstructured in human language. The challenge is twofold: to grasp the meaning and extract the relevant information to the food trends. We proposed a bag of words technique to gather relevant information in the Indonesian language for feature extracting purpose. While K-mean Clustering and Simple Additive Weighting (SAW) algorithm are proposed to draw up the food rank. In order to measure the accuracy, we compare our result with the sales data of some restaurants in Yogyakarta. We test the algorithm using 4 weeks of data, the result is compared against the available data and an accuracy of 72.75 % is achieved.