In economies where the role of big data is ever-increasing, companies turn to business intelligence (BI) and reporting tools to have their data analyzed and presented in a precise and readable format.
As a data scientist, I’m always working with companies looking for actionable insights. Both large enterprises and startups today are ready to invest in BI tools and the relevant engineering work to explore and visualize their data.
In this article, I review an open source data visualization tool Superset designed by Airbnb. We used Superset in one of our recent projects and it served us really well in most cases. In this article, I will tell why we chose Superset among other BI tools, what are the main benefits and drawbacks of the platform.
We used Superset in our project for a fitness mobile app with a huge fast-growing customer base. On the one hand, a BI tool was requested by business stakeholders, who needed a number of specific reports to monitor trend changes in application usage and better understand customer behavior. On the other hand, a BI tool could be used by our data science team to perform exploratory data analysis in relation to different user cohorts before building Machine Learning models.
We needed a tool that would satisfy the following requirements:
After searching for available solutions, we selected SupersetandPentahofor further evaluation.
Superset was seen as a more attractive tool for us for the following reasons:
Superset is a data exploration platform designed to be visual, intuitive and interactive. Superset’s main goal is to make it easy to slice, dice and visualize data. Its developer claims that Superset can perform analytics at the speed of thought. As we have already mentioned the open source data visualization tool is written in pythonic web framework Flask.
This project was originally named Panoramix, was renamed to Caravel in March 2016, and is currently named Superset as of November 2016.Source.
Superset allows integration of multiple data sources but in our project, we created a single BI data warehouse that was the only data source for Superset.
2. After you’ve added a data source you should add tables from that DB and specify fields’ properties. You can specify if a field is groupable, filterable, temporal and so on. Also, you can add custom metrics (in addition to default ones like COUNT, COUNT DISTINCT, SUM, etc.).
3. Then you create slices. Slice is a single plot based on your data. Note that you can build slices for just one table at a time. But it’s not a problem because you can always create a view by joining as many tables as needed. Your future BI report can consist of one or more slices.
Superset worked perfectly for us when we used it to plot aggregated metrics like distribution of users grouped by OS (iOS, Android), age groups, gender and so on. But the tool appeared to be helpless when we wanted to plot aggregations of a higher level.
