Monday, March 25, 2013

PayPal's Chief Data Scientist


Reading an article written by Dan Woods about PayPal’s Chief Scientist, Mok Oh. It is very interesting and I have summed up the article in this post, but if you want to read the details of the original article, please click here.

Who is Mok Oh?
Mok Oh is the Chief Scientist of PayPal. He is responsible for ensuring that PayPal makes effective use of data, exploring big data for predictive indications about future buying trends in its customers. He is also responsible for bringing PayPal users get the best possible experiences. According to Oh, the mission of the data scientist is to build a credible data model, then get more data about observed behavior to improve the model.

How to model the Commercial Brain?
PayPal is trying to model a person’s brain, at least the part of the brain that decides how to shop, when to shop, and what you want. PayPal is trying to reverse-engineer transactional data to figure out what people are going to buy next. Despite having petabytes f data, it is never enough since there are a lot of behaviors Paypal needs to model.

What important advice offered by Oh to data scientists?
Data scientists tend to focus on a very small piece of data without occasionally stepping back to take in the whole problem. The other pitfall is a tendency to over-focus on technology when business acuity is important. Organizations tend to employ the most technology-savvy person on the job, rather than the most business-savvy person. Oh says “knowing the right question, or problem to solve is very important”. It is because computer scientists are not trained to ask the right business questions.

What are key skills of the data scientist?
In the long run, the key skills of the data scientist, regardless of the amount of data he or she can collect or the type of tool he or she is master at, may be an intuitive sense of appropriate business questions as the analysis begins, and solid judgment when it reaches the end. Furthermore, the data scientist needs to look at each dataset on a case-by-case basis so as to understand the “signal strength” of data.






No comments:

Post a Comment