The process of making purchasing decisions has witnessed a noticeable change in the last two decades. Previously, consumers were making their decisions based on the products’ advertisements or input provided by the seller. Nowadays, consumers have become more reliant on opinions / online reviews posted on social media like Facebook and Twitter, e-commerce websites like Amazon, and eBay, applications platforms like Google Play, and Apple App, etc.. to make a variety of different decisions ranging from what product to buy, to what movies to watch or which hotel to reserve and the list goes on. Recent surveys state that around 90% of customers consider reviews while making a purchase, especially for big-ticket purchases or to determine their quality (Power of Reviews, 2014; Bright Local, 2020). These reviews provide a rich resource of information not only to the consumers but also to the business owners and app developers as they help them to understand the consumers/users' needs since reviews mostly contain valuable feedback that affects positively or negatively on the business reputation and on attracting new customers. For instance, people often express their feelings by leaving positive or negative feedback or by giving a hint that indicates the probability of but this product again.
The datasets used in this paper were downloaded from the Kaggle website: https://www.kaggle.com/ therealsampat/google-play-apps-reviews?select=reviews.csv. Datasets were scraped from Google PlayStore. The reason for choosing Google Play store data is because Google Play is considered one of the leading distribution platforms for applications. Based on research, it has more than 2 million apps with download features for over a hundred million every day (Statista, 2016). What distinguishes Google Play Store is the distributing facilitation for applications and the users' facilities to easily rate and post a review for apps they purchase or download.
- Language detection
- exploratory text analysis
- create a corpus converted to matrix
- Apply the technique of classification model that will help us to examine and predict the purchase behavior using a popular classification algorithm “Lasso logistic regression”.
- Emotion classification