Data Science Products: what you need to know
Justin serves as Head of Data Science at Yelp. He is a career data professional and Data Science leader with experience in multiple industries and companies. Previously, Justin was the Director of Research and Data Science at Cloudera Fast Forward Labs, head of Applied Machine Learning at Fitbit, the head of Cisco’s Enterprise Data Science Office and a Big Data Systems Engineer with Booz Allen Hamilton. In another life, Justin served as a Marine Corps Officer, with a focus in Systems Analytics and Device Intelligence.
He recently spoke at Yelp HQ in San Francisco and discussed Data Science Products, including what they are and why we should care about them.
To learn all about Data Science Products and the role of data science in product management, you should watch the whole video from Justin Norman’s Speaker Series in San Francisco. The highlights are detailed below.
Product Data Science means building data products, tools and measurement strategies that impact the consumer using rigorous statistical methodology, expertise, and experience.
Just like products have a product lifecycle, a data science product has a data science product lifecycle.
Here is the Data Science (DS) Product Lifecycle:
- Ideation and Design
- Data Exploration and Pipeline Development
- Product Development
Types of DS Products:
- Application embedded: Some kind of applied (usually probabilistic) data science or Machine Learning built into an application as a feature. Ideally, without user knowledge.
- Actionable research: It’s not a feature. It’s not something that’s going into the application directly. It’s metrics to help make decisions on where time and resources should be spent.
- Experiment design: Have an experiment plan to help guide the conversation between Data Scientists and Product Managers about the measurement strategy. The Data Scientists, Engineers, and PMs must partner during the experiment execution. Your plan must keep up with the changes in thinking that come as a product evolves.
- Reusable analysis tools: Reusable code and tools enable and promote the use of data in product development. Data Science helps Product Managers explore and select which metrics to use.
- Confident monitoring: When you release a product that has some kind of treatment from a statistical perspective in it, you can pretty much guarantee that you’re gonna have to redo it soon. So, if you have monitoring in place you’re able to react to that quickly.
- External reporting: These data science products promote the brand and engage the community outside of the core product.
The DS Product Manager Toolbox
- Data Lifecycle and Pipeline Management
- Experimentation and metrics
- DS Development Process
Core Data Science PM Responsibilities:
- Decide on core function, audience and desired use of Data Science product
- Evaluate any inputs and ensure they’re maintained through the lifecycle
- Orchestrate the cross-functional team (Data Engineering, Research, Applied Machine Learning, Data Science, Machine Learning Engineering, Software Engineering, Data Platform, Metrics Platform, Experimentation Platform)
- Decide on key interfaces and designs: data features and UI/X
- Work with Engineers and Data Scientists during development to determine the tech stack
- Coordinate maintenance and support after the product release