For the things you don’t already know – there’s an app for that

2 min read

I thought it may be cool to check out what some app developers out there are coming up with. The Prior Knowledge app seemed to be a good one to take a closer look at.

Knowledge is king, and we all have a little part of us that wishes we could predict the future. Well, mystical fortune tellers aside; what if there was a way?

Prior Knowledge runs Veritable – a system expertly designed for making pretty good predictions based on tabular data. Whether it’s predictive time tracking, product recommendations, purchase predictions or disease predictions you’re after, this is a great solution to getting your hands on the information you don’t have.

“With the release of Veritable, every developer can now be a data scientist” says CEO and co-founder Eric Jonas.

What makes Veritable stand out from the prediction engine crowd is that it learns a “joint model” of the data. This means it can make predictions about any combination of columns in an analysis, given values for any other combination of columns.

Also, it isn’t swayed by missing data. From incomplete web profiles to missing answers in survey results, the issue of missing data is bound to come up. Unfortunately, many statistical and machine learning approaches expect a complete table, which normally results in an individual guessing the missing values or just throwing the incomplete rows out; neither of which is helpful in achieving accurate results. Veritable, however, has a unique way of “filling in the blanks”, considering every possible value in that space and weighing it up accordingly. A much better option. The Prior Knowledge blog covers a little more about these benefits.

In a nutshell, Prior Knowledge’s Veritable is a predictive database that finds the causes behind data and uses these causes to predict and explain the future in an easy to use and highly effective way. A pretty cool addition to the world of app development.

Veritable is currently accessible through a RESTful HTTP interface, and via Ruby and Python client libraries.

Contact


Want to work with me or the Lab19 team? Great news! Reach me using the relevant links below: