Getting started in Data analytics

By Stephen Langley, Head of Data Analytics Group, Securities and Futures Commission

Stephen Langley, Head of Data Analytics Group, Securities and Futures Commission

Although data science has been around 50 years or so it is only quite recently that it has become realistically and economically viable for smaller organisations to use. Low cost powerful computers (or rented compute power), vast amounts of data (which drives accuracy in Artificial Intelligence (AI), and decades of mathematical and statistical research readily available in easy to use software libraries have finally made the promise of advanced analytics viable.

If you are thinking about getting into data analytics I want to offer some practical advice on how to get started and things to think about. In particular I will talk about what I think are some of the foundations for analytics: culture, people, and data. These may not be the first things people think about, but in my experience they are crucial.

Culture

It is important that you understand the culture of your organisation: Is it interested in and supportive of using technology to move forward or is it conservative and resistant to change? Are senior leaders trying to drive change but hampered by the middle ranks less keen to change? The important thing is to understand your culture so that you can implement change such as new analytics programs at a pace that can be absorbed by the organisation. Understand the appetite for change and do not try to force it.

It is also important to understand where you are on the ‘analytics maturity curve’ which starts with basic reporting and goes through business intelligence all the way to AI. Understand that it is not realistic to go straight from basic reporting to AI in one go.

People

There are three different groups of people that need to be managed and communicated with—but in different ways.

First are senior leaders. Typically they will not be technical so communications need to be in simple jargon free language and focused on the business deliverables. You will likely need to provide some executive education of what analytics actually is and what business benefits it can deliver before you can even have a conversation about analytics projects.

Once you start to work with data from multiple systems you typically find inconsistent security access issues and that no one has responsibility for data quality

Next is the middle management level. You should work to raise their awareness of how analytics can help them do their jobs and assuage their fears of being monitored or replaced. Using the term Augmented Intelligence rather than Artificial Intelligence is helpful and probably more accurate. Look for people already working in some areas of analytics in your departments and support them and turn them into champions for analytics projects.

Finally there is the IT Department. Analytics projects are often sponsored by business units or analytics teams and can be very different from IT projects. IT projects typically have clear requirements and clear deliverables and so, can be made into a defined project fairly easily. In contrast Analytics problems may be high level with unclear requirements and deliverables requiring a lot of research and testing, all of which require a more flexible timeline. For this reason IT projects and Analytics projects need to be handled differently. The important thing is to ensure there is clarity on how these different types of projects are handled.

Data

Last, but by no means least, is data.

In the past year I have spoken to many data scientists and all agree on one thing—that more than 50 percent of the work effort is spent in data cleansing and data quality activities.

Historically in corporate environments the focus is on individual systems but analytics increasingly requires data from multiple systems—internal and external. Once you start to work with data from multiple systems you typically find inconsistent security access issues and that no one has responsibility for data quality.

Data standards, data governance and data lifecycle may be unfamiliar terms in some organisations. The more you do analytics the more you will become familiar with them.

Do not underestimate the amount of work effort required to collect, cleanse and quality check your data in order for it to be a solid foundation for your analytics efforts.

These are some important topics to think about as you get into analytics, and topics you should think about before you think about technologies.