How it all started
A few years ago, we established a Data Science department in Statnett. Being the transmission system operator of the Norwegian power system, Statnett operates approximately 11,000 km of power lines and cables, with about 150 substations. Every single day we gather vast amounts of data from our assets in the power system, and our goal is to enable more data-driven decisions.
We have made progress working together with core departments in Statnett to develop tools and machine learning models that solve advanced challenges. But we quickly realized that to become a more data-driven organization we needed to develop new skills throughout the company.
Luckily, we had a lot of colleagues who wanted to learn. Our Python classes were full, people attended our lectures and tested their skills on code kata’s.
But that wasn’t enough. A 2-3-hour course might be inspiring, but to truly be able to use data for daily analytical tasks our colleagues asked for more training. An idea was born.
Developing relevant data science skills
We could have chosen another strategy – to concentrate the data skills around the Data Science department. Issues around data quality and securing the right interpretation of large data sets would perhaps be more easily handled that way, but we truly believed in a decentralized system. By installing the right skills and tools in the core business we could get more value and scale faster.
We set an ambitious goal. A group of employees in Statnett would attend a full time Data Science academy for three months. They would have to leave their regular tasks behind and be fully commited to the classes.
The HR department supported the whole process, including the selection of employees. The application was open for everyone, but they needed support from their leaders. This was an important aspect, as we wanted the department of the attendee to commit to the program as well. There is no use in learning new skills if you are expected to go back to the old habits after the program!
From Python to data wrangling
The content of the course was developed by us, and tightly linked to Statnett’s core business. That way we ensured a more interesting academy for our experienced colleagues. It quickly became clear to them how they could use data science tools in their daily work routine.
The academy was a mix of coding sessions, lectures followed by workshops, self-studies and group work. We focused on a group of important skills:
- Basic programming, especially focusing on Python
- Learning about and getting access to data sources, both inside Statnett and external
- Data visualization and communicating results of data analysis
- Data prep/data wrangling – the process of transforming raw data and gaining control over missing data
The classes were taught by employees in the Data Science department, in addition to shorter workshops and lectures by colleagues from across the company.
What we learned
Our goal was to utilize Statnett’s own data in all case studies and tasks. But at the beginning of the course we realized that some of the attendees needed more basic coding and data visualization skills, and we adjusted the progression. We focused more on the basic skills and waited till the end with the more advanced use cases from the organization.
This is one of the reasons why we believe it’s fundamental to build data skills through an internally developed program. We were able to evaluate consecutively and adjusted the program to ensure the right outcome.
We still haven’t decided what the next academy will be like, but most certainly we will start with the basic skills and build on that. A possible way forward is to admit more employees from the departments that have started using the data science tools to the next academy, to further build on what’s working.
Long term effects of learning new skills
Our attendees are now able to use data science tools to solve their day to day tasks more effectively, so they can focus their creativity on the more demanding tasks. Our goal is to ensure that employees across Statnett are inspired to learn more about the possibilities in coding and data visualization.
A relevant example is the vast amount of Excel workbooks in use. They may have started as an effective way to solve certain tasks, but many have grown way past their reasonable size. These tasks can be more effectively solved through Python tools where you update data across data sets all at once, always ensuring that every user knows which version is the latest. The hope is that these quite simple examples will inspire more people across the organization to learn.
Our first group of attendees are devoted to sharing their knowledge. In addition to initiating projects related to their new skillset, our ambition is that they will also involve other colleagues interested in learning. The course administrator will still be available as a resource for our alumni, helping them putting their newly acquired skills into use in their daily work.
Better collaboration between IT and core business
A side effect of the academy is that data skills make you a better product owner. When employees in core business know more about the possibilities in programming, data visualization and machine learning, they can work better together with the IT department in further developing the products and services. In fact, this is an important aspect of digital transformation. Companies all over are looking to utilize their data assets in new ways. Some of this is best solved through large scale IT projects, but a lot of the business value is gained through simple data tools for continuous improvement.
What happens to the Data Science department when our colleagues can solve their own problems using data and coding? Well, our ambitions are to further build our capacity as a leader in our field. We just started working on a research and development project with reinforcement learning, and we will keep working on better methods and smarter use of data.