Farewell 2016 – Lessons for anyone who loves working with data

by Tejas Mehta | Dec. 28, 2016

We at Parks Associates work with tremendous amount of data in an attempt to understand the complex and ever-changing landscape of digital living. A sizeable share of our work involves looking ahead to where digital technologies are heading (see our new whitepaper, Top 10 Consumer IoT Trends in 2017) as we help our clients stay ahead of the curve. However, as we bid farewell to 2016, an eventful year, I stepped back and looked at some of the broader news stories outside the realm of technology and relate it to how it impacts the work we do here at Parks. Here are my top 3 takeaways:

1. Data comes before science in "data science."

2016 had its fair share of elections and referendums around the world including the presidential elections in the US, the Brexit vote across the pond, and the Italian constitutional referendum to name a few.  In the aftermath of the results of the some of these polls, data science and predictive modeling as a discipline has lost some of its sheen. As a practicing data scientist, the events of 2016 has reinforced the fundamental tenet of data science which is that the integrity of data is paramount. In the context of polling and surveys, the importance of understanding and accounting for the biases in sampling and survey responses cannot be overstated. The entire research team at Parks and our consumer analytics team in particular take great pains to ensure our samples are large and representative and our question language is crafted with utmost care.  We pride ourselves on the rigor of our work and are very grateful for the trust and faith our clients continue to place in our data and counsel year after year.

2. The human element is not replaceable, at least not yet.

Another news story that caught my eye was the failure of the Michigan Unemployment Insurance Agency’s MIDAS system. MIDAS is the agency’s automated system for detecting fraud in claims.  Following reports of trouble with the system, the agency conducted a review and the report of the review was released last week. The report found that an overwhelming majority of the fraud determination cases reviewed were erroneous. For me this story is cautionary tale against over reliance on machine learning algorithms without adequate human oversight. For all the progress that machine learning algorithms have made, the human element is still vitally important for the success of any data science endeavor. At Parks Associates, we continue to differentiate ourselves on the basis of the wealth of expertise we have collectively accumulated over the course of time and combine this expertise with the primary and secondary data we collect to connect the dots and deliver value to our clients.

3. Data is interesting and fun!

In keeping with the spirit of the season, I will end this note with a story that is very cheerful. In October, Oxford University Press announced that it will add Christopher Marlowe as a co-author in William Shakespeare’s Henry VI plays. Authorship of some of Shakespeare’s works (or at least sole authorship) has been a subject of speculation for a long time, but the debate was finally settled using pattern recognition and text mining algorithms. This was one of the most fun application of data science that I have seen in recent times. 

Season’s Greetings!

Further Reading:



Next: LG’s Hub takes on Amazon Echo and home robotics market
Previous: Amazon Sells Out of Echo and Tap

Comments

    Be the first to leave a comment.

Post a Comment

Have a comment? Login or create an account to start a discussion.


ParksPerspecQuote_TechSupport-PSamuels-01.png

ParksPerspecQuote_TechSupport-PSamuels-01.png

QUOTE_Parks-Perspectives-2014-JB1.jpg

QUOTE_Parks-Perspectives-2014-JB1.jpg

ParksPerspecQuote_2.png

ParksPerspecQuote_2.png