Datameer, like a lot of companies, is trying to make its data-analytics platform accessible to people who are not data scientists. It’s part of a growing trend to ease the complexity of data analytics so the practice is accessible to the far larger group of people who are more versed in using spreadsheets to do their work.
Datameer 3.0 features what the company calls “Smart Analytics,” which simplifies Hadoop by building in the work that data scientists calculate themselves. It’s time for this shift. Hadoop has achieved deep enough adoption that it should be more accessible.
A challenge is data integration — finding ways to collect all the data and clean it up so it can be analyzed. Then there is the task of actually setting up the server clusters. The user interface has to be something a non-technical person can understand. Lastly comes the analytics itself so people can make something that is relevant and useful.
Datameer 3.0 Smart Analytics attempts so simplify the data analytics in four ways:
- Clustering the records so the customer can see the relevant groups in a data set. This might be location data, phone records, operating systems, etc. It gives people the capability to look at a customer list and see how it is segmented.
- Decision trees display desired results and then, through analysis of different factors, help determine what leads customers to making a decision.
- Column dependencies show the relationships between the different columns of data so combinations can be seen that might not be obvious. This might mean showing a correlation between location and disease type or job title and credit score.
- The ability to offer a recommendation engine by automatically predicting if and how likely it is, a person will be interested in something based on historical data from many users. This allows businesses to serve up more relevant content, products or service recommendations without needing a data scientist.
CEO Stefan Groschupf put it well in a blog post today about the state of the market. Hadoop has evolved much like other technologies have, and it is no longer necessary to build something from scratch. It’s possible to buy services that are not custom-made and, as a result, difficult to maintain over time. Instead, it’s time for more services that make the promise of Hadoop a reality.