Reaction Prediction: The Case of Tweets from Luxury Fashion Brands

Abstract

Social media platforms represent an essential tool for both consumers and marketers. Meanwhile, luxury fashion brands play a key role in fashion, one of the most important industries of the world economy. Despite assumptions to the contrary, social media platforms and luxury fashion brands do mix, especially in the recent time. Consequently, it is worth asking whether it is possible to predict the reaction a post will generate in the audience of luxury fashion brands. This new question is the one this thesis intends to answer. To do so, the concept of reaction is defined through a novel composite index that is created and named Tweet reaction overall score (TROS), which is one of the solid and relevant contributions this thesis makes. Then, several predictive models are implemented, based on a wide range of different learning algorithms. The results show that it is indeed possible to predict the TROS that a post on Twitter will obtain in the audience of luxury fashion brands the day it is posted.