Domo + Jupyter for the Window Into App Sentiment
Using Domo and Jupyter notebooks to model user sentiment for Android Applications The post Domo + Jupyter for the Window Into App Sentiment first appeared on Blog.
Earlier this year, one of our users made a post about how she used Domo and Jupyter Notebooks to run emotion detection, a type of sentiment analysis, on app reviews from the Android App Store.
We decided that we wanted to share with everybody some code on how this works. For this analysis, we collect reviews and attributes from some of our favorite food delivery apps, combine the information into a dataset, and create a dashboard showing the sentiment analysis of each review.
Kendall Ruber shared her code with us, and we have recreated a dashboard below in Domo using our Jupyter Notebook integration. Our Jupyter integration allows for more advanced analysis techniques and models to be developed and deployed entirely within Domo with ease.
For instance, instead of doing something out of the box such as a word cloud to try and figure out what sentiment is (which can be different from the App Store Ratings), we can get a better idea of our users’ sentiment by employing Python’s Natural Language Processing methods within Domo. We run the reviews through an emotion detection model, one Kendall found on Hugging Face, and with minimal effort we have an ML model developed and deployed.
We also can take advantage of the scheduling options within Domo. For instance, this Dashboard below will update daily at 09:00 UTC. Other options are also available, and current Domo users will find the options familiar.
Additionally, we bring in data from the Google Play Store API through python code written in Jupyter. This puts data from 3rd party APIs that don’t yet have a Domo Connector within reach of your Domo environment.
Finally, we also now support sharing and collaborating on notebooks, as well as integration with Accounts within Domo.
Dashboard below, code and instructions found on our GitHub site.