Study of Automotive Brands Popularity in Indonesia Using Twitter Data

Abstract

Social media has been a widespread phenomenon in the recent years. People shared a
lot of thought in social media, and these data posted on the internet could be used for study and
researches. As one of the fastest growing social network, Twitter is a particularly popular social
media to be studied because it allows researchers to access their data. However, this data
haven’t been utilized in many researches in Indonesia. Being one of the biggest industries in
Indonesia, automotive industry is an interesting subject to study. A wide range of people buys
vehicles, and even gather as communities based on their car or motorcycle brand preference.
This research will look the correlation between Twitter chatter of automotive brands and the
sales of the brands in Indonesia. The tweets that will be analyzed are tweets from April and May
2015. Factors such as sentiment and tweet rate are expected to be able to predict the popularity
of a brand. In this research, keywords will be defined to capture tweets using Streaming API.
The tweets will be cleaned up and then processed to view the popularity score. The tweets’
sentiment will also be analyzed to help improving accuracy of correlation with real world. The
Twitter results of sentiment analysis and tweet rate will be compared with real world sales
results published by GAIKINDO and AISI. However, the results of the experiment shows a low
correlation between Twitter data and real world popularity.
Sentiment Analysis, Data Mining, Twitter Mining, Automotive Industry