Predicting user engagement with textual, visual, and social media features for online travel agencies' Instagram post: evidence from machine learning

ABSTRACT

By utilizing supervised, unsupervised, and transfer learning techniques, the present article analyzes the entire three major online travel agencies’ Instagram posts (n = 6,083) to investigate which features contribute more to predicting the user engagement. Among 109 textual, visual, and social media post specific features that we initially extracted, we find the important features using the XGBoost algorithm and estimate the effects of each feature on user engagement (i.e. number of likes) using Negative Binomial regression. The results indicate that OTAs should emphasize the travel related emotion, luxurious, outdoorsy, and celebration in the post wordings in captions but should avoid the big words (words with more than six letters). In terms of images, it is recommended to use the image with fewer lines, fewer parallel lines, but more corners. For an Instagram message-delivering strategy, uploading a post during the evening is recommended.

KEYWORDS:

• Transfer learning
• Machine learning
• Instagram
• Online travel shopping
• Online travel agency (OTA)

Disclosure statement

No potential conflict of interest was reported by the author(s).

Correction Statement

This article has been corrected with minor changes. These changes do not impact the academic content of the article.