In The Trenches With Data Scientist Andrea Leistra
August 30, 2016
In this edition, Andrea Leistra takes us through a typical day in the life of a SocialCode data scientist.
Hometown: Sioux Falls, South Dakota
Current home: Silver Spring, Maryland
Education: PhD in Astronomy from the University of Arizona
Time At SocialCode: 8 months
Surprising fact about you: I nearly walked into a kangaroo when I spent a week in Australia observing a telescope for my PhD thesis.
The greatest musician of all time is: Amy Ray
In your free time, we will find you: I enjoy cooking, and I can make a mean saag paneer.
Favorite SocialCode office snack: Coffee.
What does a Data Scientist do at SocialCode and how does it impact the business?
A data scientist identifies common features and metrics that drive the best performing ad campaigns. We provide key insights to help the client team show the right ads to the right audiences. One of our responsibilities is to identify what aspects of creatives influence performance. We identify common elements of winning creatives and determine which features are worth testing. This allows the client team to run the right creative tests to successfully optimize their ad campaigns.
What can you do here that you can’t elsewhere?
SocialCode has a variety of datasets that can be interpreted in many different ways. Our Solutions team runs a wide range of sophisticated ads campaigns on behalf of a diverse client base, which provides me and my fellow data scientists with the opportunity to analyze diverse datasets. Over the past eight months, I’ve done everything from analyzing the formality of language in an ad to assessing the types of images that drive the highest engagement.
How have you grown as a professional at SocialCode?
I’ve learned how to use image labelling to identify winning creatives in ads campaigns. Image labelling involves extracting meaningful information, such as text and language style, from digital images. Different API tools extract unique aspects of a digital image. I’ve had the opportunity to assess the advantages and disadvantages of various image labelling APIs, and to select the best API integrations for SocialCode.
What’s the most exciting challenge you solved at SocialCode?
For one of our CPG clients, I found a method to automatically label creative images into taxonomic groups. The existing software lacked the capability to detect that both cats and dogs belong in a larger category of animals. I found an API tool without this limitation, and was able to develop a functionality that can detect that a cat, dog, and a horse belong to the hierarchy of animals. This image labelling functionality provided actionable insights to the client team on what aspects of creatives delivered the highest audience engagement.
Does a career at SocialCode sound like something for you? Check out our openings.