Graph databases team up with BI: It’s all about relationships
Constructing the same query using SQL (and a relational database management system) would take much longer because each vehicle has so many parts, and there may be hundreds of suppliers involved. Moreover, the relationships among them are not identified in a row and column format. “TigerGraph already knows which orders are connected to which features, such as A/C or heated seats,” explained Deshpande. “The sequence of going from orders to features, features to parts, and parts to suppliers is already determined. It is also possible to determine which parts are shared across different vehicle models, which allows parts to be reallocated if needed.” The combination of a data warehouse for storage, TigerGraph for analytics, and Tableau for reporting gives users access to a wide range of enterprise data sources, a strong analytics engine, and a familiar interface.
Associative engines make connections
Not all software products that show relationships are graph-based. The broader category of “associative technology” also focuses on relationships rather than rows and columns. Qlik is an example of such a vendor. The company began as the provider of a product selection tool more than 20 years ago. That software tool helped purchasers decide what combination of features fit best for a particular use. Its associative engine was then broadened to include analytics in other domains. Associative relationships are indexed and stored, which reduces the time for analyses.
“Our visual interface lets users see what is related or unrelated to the content of the query,” says Josh Good, VP of product marketing for data analytics at Qlik. “They can also see what is close but unrelated, and become aware of what they might miss with a traditional query.” The absence of an expected association can be informative. A leading global bank and investment firm discovered nearly $20 million in its mortgage pipeline that was not associated with any particular loan processor. The firm immediately pursued these cases, generating tremendous value from a single discovery that would not have been possible with query-based tools.
Qlik’s products include Qlik Sense, which provides the visualization and data analysis capabilities; Qlik Catalog, for locating and preparing data; and the Qlik Data Integration Platform, which incorporates multiple products acquired from Attunity in early 2019. The platform also includes Qlik Replicate, which accelerates data replication, ingestion, and streaming from a variety of sources, and Qlik Compose, which provides automated data ingestion and transformation of data lakes and data warehouses to get its data ready for analysis.
“Many of our customers are healthcare companies using Qlik to improve medical outcomes and contain costs,” said Good. “In one case, a healthcare company discovered that a certain type of surgical glue had the same benefits whether it was the brand name or the generic version, allowing cost reductions while keeping the same level of positive outcome.” Although this question could have been asked in a query, the user would have had to construct the data model and the query in advance. With Qlik, the user automatically sees the relationships without having to ask that particular question. In this case, the user could see that different glues are selected by different physicians, and can see with one click that the medical outcomes are equivalent.
Data models and queries are less expensive to build in Qlik, according to Good. “In a traditional query, you have to build it in advance, the process is complex, and the cost is high, so people ask fewer questions. If it is easy to ask ad hoc questions and receive an answer, you are likely to do it more often, which leads to more frequent insights.” Unexpected insights can also be achieved, or inconsistencies detected that point to a problem that may not have been recognized beforehand.
Data models and queries are less expensive to build in Qlik, according to Good. “In a traditional query, you have to build it in advance, the process is complex, and the cost is high, so people ask fewer questions. If it is easy to ask ad hoc questions and receive an answer, you are likely to do it more often, which leads to more frequent insights.” Unexpected insights can also be achieved, or inconsistencies detected that point to a problem that may not have been recognized beforehand.