Empowering the Citizen Data Scientist

Intel Corporation explores how the role of data scientist has evolved beyond analysis, and what IT and business leaders need to do to establish the right skills across the organization to ensure advanced analytics success.

 

Whatever industry you’re in, you should be thinking about how you’re growing stock of data could provide better insights for smarter decision making.

Doing this well requires a complex and specialized skill set. Companies trying to move away from traditional business intelligence (BI) and onto more complex forms of analytics will find it impossible without strong mathematical capabilities. As their data grows in size and complexity, many must invest in upskilling or hiring new people.

The key difference between the data scientist of today and the data analyst of yesterday is in the asking and answering of questions. Previously, a business user would ask an analyst: “how many widgets did we sell last month?”. The analyst would then run queries on the company’s relational database and create a report answering that question. Now, data scientists ask the questions. Business user have higher-level requests, such as “how can we better understand our customers?” and the data scientist figures out what questions need to be asked of what data, where that data sits, and how to bring it all together and model it to generate insights. And they’re not just mining structured databases anymore – they’re looking at social media, recordings of call center conversations, and any other type of information the business collects daily. They might also be getting richer insights by using machine learning and artificial intelligence (AI) to tease out nuances that wouldn’t have been captured before (learn more here about the evolution of analytics and how you can make the most of it).

At the same time, business units (BUs) are also getting closer to the data. In many organizations, once a data scientist has worked out the data strategy and the analytics algorithms to be applied, business users themselves can use off-the-shelf applications to ask the questions.

Empower the empowerers

I often speak to customers about how to use analytics to drive business insight. Let’s consider the analytics experts first – those who empower all their colleagues by building foundational data strategies and algorithms. You may well already have analysts or data scientists on staff, but some may have statistics or mathematical backgrounds, while others may have computer science or more specific SQL training. By looking at the types of questions they’re already able to answer, you can map the holes to plug to help you move towards greater analytics sophistication. For example, if better forecasting is an important business need, then you’ll need someone with the ability to calculate prediction accuracy, which means they need to be strong on statistics and probability. It’s possible to foster these skills in your existing team by providing them with access to the plethora of tools and training opportunities available.

An area that often requires a new hire is dealing with unstructured data, as this is the biggest change for those used to dealing only with structured databases. Having real-world, specific experience of handling this sort of data, identifying and extracting the usable nuggets, is invaluable.

Encourage an analytics-driven culture

When it comes to the BU users, it’s not just about giving them access to an application – it’s important they understand the different types of analytics (such as predictive, prescriptive and cognitive), which type(s) are available to them, and what value they deliver.

This can be a challenge. How do you manage changes that could ruffle feathers? For seasoned sales people used to going with their gut instinct, trusting the data instead might not be the most welcome idea. To succeed, change like this needs to be driven from the top down. This is an approach that we took at Intel when making analytics-based changes across areas like product forecasting, pricing, and manufacturing.

Some companies are taking proactive and exciting steps to foster an analytics-focused culture. For example, Chevron supplements its existing data science skills by running an annual competition to identify hidden analytics talent among its workforce. In doing so, it has built a celebrated team of ‘analytics rock stars'.

Takeaway tips

While it’s critical to think carefully about your data science upskilling strategy, day-to-day tactics are also important. If you or your teams are struggling to get the answers you need from your data, try following these steps (explained further here):

  1. Ask a more focused question: If asking the big questions isn’t working, try focusing on one area or detail and probing it with some more targeted questioning. You can then extrapolate to understand the bigger picture.
  2. Improve your algorithm: Creating the right algorithm is an iterative process. Tweak the parameters or choose a different model and see what difference it makes.
  3. Clean up your data: You’re never going to gain game-changing insights from dirty data.
  4. Use different data: If you’re using the same old data sources, you’re likely to get the same sorts of outcomes. Try adding a new source to help bring in new perspectives.

It’s an exciting time to be involved in analytics and data science. New technologies are enabling more complex analytics, and the inclusion of the citizen data scientist means new perspectives, new questions, and new insights are there to be uncovered. You can learn more about how to start your company’s journey to the next level of analytics by reading this ‘Getting Started with Analytics’ Planning Guide.