Roomba Reviews

RStudio, Text Analysis, Web Scraping, Topic Models, Data Wrangling, Review Analysis, Clustering Algorithms

I performed a text analysis on reviews for the Roomba 650 and the Roomba 880. After tidying up the data frame and removing common words, I visualized the top ten words used for both products to see if there were any common trends or differentiation between the two products. After running my analysis, both Roombas shared many of the same frequent words related to cleaning around the house.

I created sentiment scores for each product. The sentiment score was calculated by the amount of positive words minus negative words. The visualization of the sentiment scores show that the Roomba 880 reviews had a better positive sentiment than the Roomba 650.

I then created a topic model to see if there are clusters of words in the data containing our reviews. After running the topic model and visualizing the cluster of words that appeared in various numbers of topics I came to the conclusion that three topics would be sufficient for analysis.

The graph below shows the frequent word clusters that came across in the review data. The three overall topics I got from the reviews were: Topic 1 - the benefits of a Roomba, Topic 2 - focused on the Roomba 650, and Topic 3 - focused on the Roomba 850.

Overall, I found that reviews for both Roomba products discussed how the products were great for cleaning around the house, saving time, and picking up difficult objects such as hair. However, with the sentiment score I developed, it appears that the Roomba 880 reviews have a more positive sentiment compared to the Roomba 650.