Data Points

Over the last few years' emphasis of the Strata Conference has shifted from a pure technology conference to a data science and machine learning conference emphasizing the role that data plays in the practice of data science and the technology that is used to make data useful. 

The questions that came up the most at our booth last week circled 3 use cases:

1. Many visitors to our booth wanted to learn more about our big data/ cloud technology that could connect their data to geospatial data such as the risk, demographics and live weather we exhibited. Many brought up projects they are doing using cloud-based data science products to crunch through really large data sets and the feedback was very positive around the possibilities of bringing location to their data science models.

 

2. The next most common question was around processing and attributing "what's near" our geolocation data. We have mobile or sensor data, but not a lot of knowledge around it outside what the device itself is providing.  Can we identify nearest business, demographics of the area, etc. How fresh is the business data?

 

3. Several people told us about their teams or organizations experiences building their own geofences. Some have sourced or are evaluating vendors after discovering the complexities that exist in a) building accurate geofences and b) keeping them up to date. For unique use cases building it is their only option, regardless of the challenges. For others, halting the project to evaluate a buy versus build became a necessity when realizing the significant upfront time required. 

Are any of your seeing similar challenges? 

Any other key learnings from those that attended Strata?

------------------------------
Colin Mattison
Solution Consultant - Data
Pitney Bowes
------------------------------

Thanks for kicking this off Colin.

I attended Strata on Wednesday only. I did not spend a lot of time at the PB booth, just because the good people there were wall-to-wall busy with visitors! I did catch a number of workshops though and one thing that jumped out at me was the emphasis on data quality.

AI and machine learning are the hot topics, but in two separate workshops I attended speakers emphasized a "walk before you run" approach. One session was about democratization of data science, another about building a data-assisted organization. Each of the speakers showed the Data Science Hierarchy of Needs (or their modified version of it anyway). Each spent time making the point that the yeoman's work of cleaning and organizing comes first - a necessary precondition to doing the high-profile AI work.

I expect this is old news to a lot of folks in this community. But it was interesting to me to better understand the role that tools like our Spectrum Technology Platform plays in supporting data science.

------------------------------
David Fondacaro
PITNEY BOWES SOFTWARE, INC
Troy NY
------------------------------