Logistics and Domestic Delivery Services Performance in Covid-19 Era: A Sentiment Analysis Approach

Abstract

The Covid-19 pandemic has hurt the business sector as badly as the health sector, including the logistics sector and delivery services. This sector is entering a period of uncertainty both in terms of government policies and demand which can affect the performance of the services provided. Evaluation needs to be done to see whether the logistics and shipping services sector, especially in Indonesia, is ready to face a pandemic. The availability of official service accounts of each service provider on Twitter social media as a forum for complaints and public aspirations can be used to evaluate service performance by measuring customer satisfaction levels through sentiment analysis before and during the pandemic. Various kinds of research on sentiment analysis have been carried out, but the time window sentiment analysis especially Time-Window Lexicon TF-IDF SVM integrating model approach has not been widely used. The data were obtained by scrapping the entire data for October 2019 until September 2020. ± 10.000 random stratified training samples data per month per service provider were taken and labeled using lexical approach for classification model creation The classification model was done with 89,01% accuracy, which is then deployed to predict the sentiment label of the whole data. This study provides the results that: (1) the Covid-19 pandemic significantly increase the number of tweet that indicate more people use the service, (2) the Covid-19 pandemic have decreased the performance of logistics and delivery services especially at the first three months of the pandemic period, and (3) the most frequent negative opinion that significantly affects service performance is late in delivery.