Advanced data visualization guide: How visualizing data can boost BI
A comprehensive collection of articles, videos and more, hand-picked by our editors
Data visualization software is all the rage today. Many businesses view the technology as a simple way to get started with advanced analytics. However, they may not be right for every application.
DonorsChoose, a New York City-based nonprofit that tracks and analyzes data on financial donations to local schools, found this out when it tried putting together a report that school administrators, local lawmakers and journalists would be able to use. At first blush, a data visualization system may seem like the best bet for this non-tech savvy bunch. But Vlad Dubovskiy, lead data scientist with the organization, said traditional visualization software would have been too limiting.
Dubovskiy said he looked at visualization software from Yellowfin Business Intelligence, GoodData Co. and Tableau Software. But he eventually settled on a system from Looker Data Sciences. The main differentiator, he said, was the ability to run advanced reports in a customizable way. "We've been able to program really sophisticated business rules," he said.
All of this isn't to say that more traditional data visualization software systems can't be customized or used to run advanced analytics. But Dubovskiy said organizations that lack a data scientist may need to rely more heavily on the vendor to set up reports. Since DonorsChoose has a data scientist on staff, it was easier for them to look beyond self-service visualization tools.
Dubovskiy said the Looker system does produce graphic visualizations of data, but this is not the main vehicle for data discovery, as it is with today's more common data visualization systems. Instead, it relies on a proprietary language called LookerML to define reporting criteria. He said this increases his control and allows him to generate relevant reports.
Most recently, DonorsChoose used the Looker system to put out its Giving Index, which breaks down charitable giving to educational organizations -- whether local school districts or colleges and universities -- by a number of characteristics. For example, the report shows that Manhattan, Brooklyn and Chicago were the localities that received the most money through charitable giving. Lower grade levels tended to receive more money, as did programs focusing on science, technology, engineering and math.
DonorsChoose has been collecting this data since it was founded in 2000. But Dubovskiy said they didn't have a good means for distributing it to schools and districts, which could then use it to better target fundraising campaigns. Initially DonorsChoose just opened up its database. But only the most tech-savvy users could find any meaningful correlations. This is what led the organization to implement the reporting system in 2013.
The application DonorsChoose implemented does share some similarities with data visualization software. It blurs the line between big data and more traditional business intelligence reporting systems. Dubovskiy said you could call it a big data system, but ultimately that's not the problem it tries to solve. The same has been said for today's visualization software systems: They can pull insights from data, but have a smaller focus than most definitions for big data describe.
All of which means there are applications to fit every level of technical expertise. Data visualization software may fit businesses that are new to analytics, while something like Looker might be better for midsize organizations like DonorsChoose that have a data scientist on team. Either way, Dubovskiy said the important thing is to just get started and fail fast.
"Try not to be a perfectionist," he said. "Just do the basics. Invite the rest of the organization and continuously improve."