Join Doug Rose for an in-depth discussion in this video Add to organizational knowledge, part of Learning Data Science: Using Agile Methodology.
- You've seen that a predictable way to make great discoveries is to allow your team to explore the data. They should have the freedom to look for interesting connections. There's also the data science life cycle that provides a structure for the team to tell interesting stories. The DSLC gives the organization visibility into what the team is doing. Instead of planning they're getting predictable delivery. That's a key way to keep your team focused on building knowledge. You're not giving them an objective.
Instead you're giving them a predictable framework. They have a weekly rhythm to share their stories. If the rest of the organization doesn't like their stories, they can always encourage the team to go in a different direction. The DSLC sprints and exploration all work together to provide insights in learning. If you're working on a data science team, you should try to keep these three things in balance. The DSLC gives the team a blueprint for how to deliver real value. The team should identify the roles and work in a cycle of questioning, researching and storytelling.
The sprints give the organization a predictable pace. Without a sprint the team is in danger of spending too much time preparing instead of delivering. Most of the time your data analyst is scrubbing large data sets. A two-week sprint forces them to work in the smallest possible chunks and encourage them to deliver smaller reports instead of one large visualization. Finally the organization needs to put a lot of emphasis on exploring the data. The team should have the freedom to follow up on interesting discoveries.
The rest of the organization will have visibility into a team's work but that work might change. The team might pivot based on a new serendipitous discovery. The balance between sprints and exploration helps keep the conversation alive. The team has the extra freedom and in return the organization gets frequent feedback. If it's done well the data science team will work closely with the organization to help learn about the business and their customer. It's a balance between light-weight structure and frequent delivery.
That being said it's not an easy balance. Sometimes the team might struggle to deliver anything interesting. Other times the data sets will seem so large and complex that it couldn't possibly be broken down into a two-week spread. This framework isn't designed to solve those problems. Instead it's just a way to shine light onto the challenge. It forces the team to think small. One way to make sure your sprint is always delivering value is to end each storytelling session with a clear call to action.
Your audience should be interested in adding to their organizational knowledge. You can help highlight the value to the organization by making clear suggestions on how they can leverage this new data. Let's imagine our running shoe website shows the clear connection between whether a shoe has a customer rating and how well it sells. In your storytelling session you should set up a clear visualization showing the connection between the sales and the ratings. They found that shoes that have no ratings are less likely to sell. Yet that shouldn't be the title of your story.
Instead you should show how your organization gains from this insight. You want the title on the visualization to be Increasing How Many of our Products are Rated Should Increase our Overall Sales. With this title you're not just saying what the organization knows but what you're doing is clearly outlining the business value. In one sprint the data science team has made a suggestion on how to increase sales. There's a call to action. If you want to increase sales, then you should encourage customers to rate their products. When you're working with data science teams, try to remember that your organization will view new knowledge in a very practical way.
Be sure to balance the DSLC with sprints and exploration to deliver stories each week. These stories should show new insights and have a clear call to action. When your team has a clear call to action, you're more likely to get interesting feedback. You're audience may ask you to follow up on your story or create new stories with even more insights.
This course shows how to structure your work within a two-week sprint. See how to work within a data science life cycle (DSLC)—a methodology for cycling through questions, research, and reporting every two weeks. Explore key practices to help your team break down the work so it fits within a two-week sprint. Learn how to use tools like question boards to encourage discussion and find essential questions. And most importantly, learn how to grow your team's shared knowledge and avoid common pitfalls.
- Defining data science success
- Determining project challenges and criteria for success
- Using a DSLC
- Iterating through DSLC sprints
- Creating a question board
- Breaking down your work
- Adding to organizational knowledge
- Avoiding pitfalls