Home Blogs

How to build your team’s data science skills from scratch

Part one: The importance of learning Python.

By Celine Gross
Published 20 January 2021

This is the first part of our ‘How to build your team’s data science skills from scratch’ series, a succession of blog posts offering ideas and practical tips to bring more data into your organisation.

Eighteen months ago, my colleague Chris Owen and I started on Social Finance’s internal data scientist training. Now, we have trained a dozen more data scientists across the organisation, beyond Social Finance’s core data / digital teams. It didn’t require a complex hiring process, nor sending staff on expensive external courses. We created a simple yet effective programme that can be adapted to any organisation — no matter its size, its purpose, or its level of data literacy.

But first… Why learn to code?

Social Finance is a non-profit organisation, delivering impact focused projects. Why is coding such a core part of our work?

It’s simple: knowing how to code gets us closer to the essence of the social issues we are trying to address. Coding enables us to manipulate more data and uncover complex patterns and trends — things we could not do in Excel. It is also a great time saving tool: simple tasks can be automated, and complex analysis completed in seconds. Thanks to coding, we deliver timelier, more impactful work.

Knowing how to code gets us closer to the essence of the social issues we are trying to address.

Last summer, we looked at patterns in exclusions with Cheshire West and Chester council. By joining 700 CSV files, we were able to track each child’s journey through school and other services and pinpoint those at risk of exclusion. Cheshire West and Chester are using these insights to redesign services and support at-risk pupils early.

We’ve been amazed by the opportunities that coding unlocked. We believe you will be too.

How to get started

Find motivated staff and protect their time

We issued a call to the whole organisation, asking for expressions of interest. We selected a cohort of six motivated staff and gave them one day a week to train (with flexibility during busy weeks) for three months.

Choose a good online course

We enjoyed Datacamp’s Data Scientist with Python course: it’s affordable and well designed. “Datacamp has been a great course — providing a mix of video content, practice exercises and projects to learn from. It allows individuals to work at their own pace and is a helpful resource to refer back to” says Victoria Walsh, who was recently part of our internal programme.

Create a peer support group

It’s important to bring a bit of social into an otherwise solitary online learning experience. It also keeps learners accountable, as they have someone else to check in with on their progress. We had one hour Python catch ups every Friday: 30 minutes of tutorial on a topic delivered by me or Chris, and 30 minutes of Q&A.

We also used a Slack channel to ask for help or chat about new concepts. It was great to see learners supporting and learning from one another.

Practice on a real project, as soon as possible

No need to do the whole course to use these new skills: after three months, staff were knowledgeable enough to start using Python in their work. They might need some support to grow in confidence. Encourage them, and let them start small by reproducing some Excel analysis in Python or by automating simple tasks. If more experienced staff are available, suggest they do some pair programming.

“I found doing project based learning was the best way to learn quickly and stay motivated”, says learner Sabrina Rafael. “And while project based learning can feel quite overwhelming, committing to small, measurable goals is a great way to overcome this.”

What’s next?

Once you have data science capabilities in house, have a think: what can you achieve that used to be out of reach?

For Elaine Merrins, Intelligence Analyst at the London Innovation and Improvement Alliance (LIIA), using Python will transform her job. We were lucky to welcome Elaine in our group of learners, and she’s looking forward to using her new skills to provide benchmarking analysis to London Boroughs:

“Adding Python to the LIIA data repertoire will enable us to analyse Children’s Social Care data sets in a way that would just not have been possible with tools such as Excel. It will enable us to answer key questions from data which were impossible before. Learning with Datacamp and the support of Celine and a cohort of fellow learners has been invaluable.”

I am happy to continue the discussion at celine.gross@socialfinance.org.uk. If you are working in children’s services, have a look at the Children’s Services Data Science Apprenticeship and our open-source code.

In forthcoming parts of this series we’ll take a look at visualisation, storytelling and analysis design. Follow us on Twitter @socfinuk to ensure you don’t miss out.

 

Sign up for the Social Finance newsletters

 

 

We'll keep your data secure and won't share it outside of Social Finance, ever. For more details read our privacy policy.