TIME SERIES ANALYSIS ON SOCIAL MEDIA POSTS: A STUDY ON TS RESTAURANTS IN HAWAII

Abstract

The effect of the social media on different aspects is wide. Use of social media seems obvious to manage continuous customer support, quick response, understanding the seasonality, get feedbacks from customers on the businesses and also see the effect of the promotional events and activities. Finding the significance of long-term effect of several events comes with challenges because of some factors such as seasonality, time trend and availability of customers’ timely response which hides the true signals about the effect of the event. In the present study, the objective is to explore the impact of the musical band and analyzing the seasonality trends of the customers against promotional events. In order to explore the capability for understanding the effect of promotional events, social media response of the customers on the musical band events organized by a restaurant chain (TS restaurants) is analyzed by applying different time series tools. Only SARIMAX model is found useful to get the association of such event on customers response using social media because along with seasonality it also takes exogenous factor into account. Facebook, Twitter and Yelp are used as social media platforms to collect the time series data of Restaurants’ (TS restaurants and other restaurant in the similar location). After the analysis, the significant effect of the promotional event (musical band) is found on the overall Yelp ratings of TS restaurants. The forecast is also provided for this restaurant chain with exogeneous factor which shows the seasonality with upwards trend. Although the model is built using data for TS restaurants, the method is generic and therefore applicable to any restaurant’s forecasting using exogenous event