Exercise

Revisiting Google app reviews

We're now going to leverage gradient boosting to perform sentiment analysis!

Remember the app ratings dataset from the first chapter? We'll now work with a related dataset for user reviews. This dataset contains the first 100 most relevant reviews for each app. Our goal is to build a model based on these reviews that predicts the Sentiment as being 'Positive', 'Neutral', or 'Negative'.

The dataset is available to you as reviews, and the target here is the 'Sentiment' column. Explore the data in the IPython Shell, and then select the correct statement from the options below.

Instructions

50 XP

Possible answers

Most of the reviews are 'Negative'.

There are some reviews with NaN values.

The target is balanced.