The world of data analytics is exciting because so much of its power lies undiscovered, and its potential is being explored together, globally. For those data professionals, whether data scientists or line of business analysts (which Gartner terms citizen data scientists), the future is still unwritten.
However, as the analytical world creates new data sets and the tools to exploit them, this novelty means data professionals can only gradually come to recognise when the ship needs course correction.
Many analytical processes have an unintended bias because they are built from the experience and perspective of a particular developer. This is a hard thing to guard for, and it can be harder still to even perceive if there is a problem in the first place.
In a 2017 report (Embrace Your Bias to Enable Analytics Clarity), Gartner recognised that bias can be intrinsic within analytical processes and, “inherent in the development of analytic models, data selection and the associated algorithms.”
The challenge here is that at times data science teams are not terribly diverse, as with many professions in society. Many data analysts come from similar backgrounds, and lower down the educational curve, only 15% of STEM degrees are awarded to women. As such, it’s up to organisations to make a concerted effort to create cross-functional teams with more diverse backgrounds to help create better analytical insights and processes.
Diversifying teams may sound like an issue for Human Resources, but with a little investigation, clear reasons for seeing this as an analytical talent and process issue become clear.
• Diverse teams will naturally share a better understanding of the wider customer base.
• There is evidence that, generally, diverse teams perform better. By embracing heterogeneity, creativity is sparked: Differing opinions, experiences, and data combine in an alchemical brew. It’s a little older, but McKinsey published a study with evidence to this point that’s well worth a read.
According to a recent report from LinkedIn, Global Recruiting Trends 2018: “Diversity, the popular phrase of the 1980s, became diversity and inclusion as the movement matured, and today has expanded to diversity, inclusion and belonging. Here’s why: diversity is being invited to the party, inclusion is being asked to dance, and belonging is dancing like no one’s watching.”
The report went on to say: “78% of companies prioritise diversity to improve culture and 62% do so to boost financial performance. Key forces are at play: changing demographics are diversifying our communities, shrinking talent pools for companies that don’t adapt. Growing evidence that diverse teams are more productive, more innovative, and more engaged also make it hard to ignore.”
Gartner suggested four major methods to assist an organisation in their efforts to transparently expose statistical bias, “so as to ensure real business impacts.” These methods were:
1. Use independent auditors to detect and analyse bias in digital personalities.
2. Compare the bias in data and analytic models with the business understanding of that bias.
3. Perform periodic reviews of analytics with an objective board of standards.
4. Review boards should challenge exclusive ownership of analytics to avoid a self-reinforcing cycle.
Diversity goes beyond the people in the business, and the recruiting and management processes that should ensure impartiality and inclusiveness. Organisations also need to diversify their data sources, because if they are simply leveraging data from their data lake or data warehouse, they are not capturing all of the potential data to a fuller, more rounded viewpoint.
Of course, any data professional is likely to be excited by data, and even more excited by really big data, and already pushing to have the all the resources that the budget can support, and they can handle.
Creating diversity in data, as in the workplace, involves hunting for elements that are identified as lacking. For a data professional, that’s often too much of their job, and very time consuming. Alteryx’s recent study, delivered by research firm IDC, discovered that data professionals spend 60% of their time getting to insight, but just 27% of that time is spent on actual analysis. That is out of balance.
These methods of working result in data professionals wasting 30% of their time. That equates to an average of 14 hours per week due to not being able to find, protect or prepare data. With the best will in the world, no analytic team will be able to create amazing creatively designed and robustly defended business insights as standard when they aren’t collaborating. When analysts are simply searching and preparing their data, and have little time to apply collective intelligence to it.
>See also: AWS championing diversity in technology
Diversity of data sources needs to be a quicker fix, so that a diverse team can use diverse data to create impactful insights. A team without the time to discuss and learn from each other are unable to utilise their diversity, to discuss, debate, disagree and decide.
It’s worth businesses considering the ancillary issues around cultural change, so that any changes made really result in the right kinds of differences that the business intends to see. In the realms of analytics, it’s a mix of breaking down silos and unifying the business, establishing agile processes that let the analysts analyse, not waste time. It’s about deriving more value from existing systems and data, and easily adding more without adding trouble.
In short, it’s about diversity and it’s about do-ability. It’s a genuinely exciting time to be in the analytics business. The field is still growing, learning, and changing for the better. Those who can harness that enthusiasm and shape a properly inclusive and effective data culture will be the new leaders of tomorrow.
Sourced by Nick Jewell, technology evangelist at Alteryx