Uncover new attributes of interest with Data Discovery Assistant

by Zeke Sexauer, on September 4, 2019

Narrative’s Data Discovery Assistant helps data buyers uncover new features to enrich their data sets and unlock new insights.

There’s a wealth of data available in the Narrative Data Streams Marketplace, and the organizations that can successfully leverage that data to enrich their own data sets gain a valuable competitive advantage. However, being able to identify the attributes that boost a signal among those that just add noise has so far proven challenging. To address this, Narrative built Data Discovery Assistant.

Narrative’s Data Discovery Assistant allows users to find high-value attributes suitable for a variety of data needs in an intuitive and easy-to-understand interface—whether as a marketer seeking to improve personalization, an analyst conducting competitive analysis, or as a data scientist feeding predictive models.

How it works

Data Discovery Assistant measures the relationship between data attributes using a method called pointwise mutual information (PMI) in order to quantify the strength of association between features.

Users simply upload a list of data they'd like to enrich to Data Discovery Assistant. We call this the “seed data.” The seed data is then run through a model that measures the relationship with the data in the Narrative Data Marketplace. Data attributes with a high PMI association are then presented to the user based on the strength of association and the number of occurrences.

screenshot of Discover Assistant report creation process

Say you’re a mobile food ordering service and you’d like to create a more tailored experience for your customers by suggesting restaurants and dishes based on the user’s profile and order history. For users without sufficient history buying through the app, these tailored recommendations would be impossible.

Using Data Discovery Assistant, though, you could upload a list of users who, say, have a proclivity for a particular cuisine, and receive observations that are correlated with those users as output. For example, you might find that people who have the ClassPass app on their phones are much more likely to get food delivery from restaurants serving sushi. You could then use that data to tailor the restaurant recommendations next time the user opens the app.

screenshot of Discover Assistant output report


Quickly find valuable predictive data

Uncovering new sources of predictive data is usually a process of trial and error. You have to come up with a hypothesis about a certain attribute that may correlate, locate and purchase a data set that contains that attribute, and then test to see if associations actually exist. If the association is weak, it’s back to the drawing board, again and again. With Data Discovery Assistant, you can skip that process entirely by seeing associations from all existing attributes at once, saving you time and money.

Better understand your data

You don’t know what you don’t know. By revealing new, and perhaps unconventional, associations with your data sets, Data Discovery Assistant helps you and your team learn something new about your data.

Try it

Data Discovery Assistant is available to all Narrative Acquire Standard and Enterprise customers. Try it out today!

If you’d like to learn more about how Narrative’s automated data discovery and acquisition tools can help you take control of your data strategy, reach out to your partner success manager or get in touch here.

Topics:Narrative Features