High performance analytics (HPA) can deliver a competitive advantage for organisations. But to reap the benefits, HPA has to become an intrinsic part of the way a company works, says ANDRÉ ZITZKE.
Within most organisations that have adopted an HPA solution, there exists a “champion” – someone whose idea it was, who headed up the implementation, who made it all happen, and who now guides the organisation’s strategic decisions with the results.
It’s great to have this person – in fact, this person is vital to the success of the implementation – but a problem arises when he moves on from his role. The truth is, despite the numerous strategic and competitive advantages that HPA offers, it hasn’t yet become an intrinsic part of the culture of organisations that have adopted it, and when the champion is deployed elsewhere, the ongoing use of the HPA solution is hampered.
This is why it is so important to understand the HPA lifecycle – not just for one person, but for the whole organisation. Once HPA has been implemented, it’s possible to keep reaping the benefits. But people have to know that it’s there, be aware of how it works and understand how it can inform and underpin any business decision.
This awareness and understanding can spell the difference between an isolated use case of HPA and an ongoing engagement with the solution that will result in true business competitiveness. For instance, a financial services provider might want to use HPA to assess the impact of a mining strike on the creditworthiness of the individuals in the area where the strike is happening, but if used sustainably, the HPA solution could be used to assess the ongoing growth potential, risks and viability of granting credit in such an area or other industries and areas where strikes happen.
Understanding the lifecycle
In order to embed HPA in an organisation so that it can deliver this profound level of insight, which is repeatable and delivers sustainable benefits, the four high-level elements of the HPA lifecycle need to be understood. They are:
Discovery
This is the process of understanding and articulating the business problem. Analysts should be brought in to work with the business people and IT people to translate the problem into something that IT can support. This is an ongoing process on both sides, until business and IT are happy that they understand the issue and the objectives so that improvements can ultimately be achieved.
Modelling
In this stage, a data miner or quantitative analyst will come in to do the modelling, be it pure descriptive statistics, predictive modelling or exploratory analysis. Depending on the business requirements, they will then use one or a combination of these techniques to come up with an analytical model, which is then verified back into the business. Remember that statistics cannot work in isolation from the business – the information fed in must be relevant for HPA to correctly interpret it and to obtain results that make business sense.
Deployment
These models now need to be put into production. Remember that to use a model for a once-off analysis provides value once-off and might have a short- to medium-term impact on the business. For a sustainable, long-term impact on the business, HPA needs to be put into active production, and this is where IT comes into play. They need to put the model into action so that it can run on a regular basis – from real time to monthly, depending on the needs of the business.
Monitoring
This is where the actual reporting happens around the model to assess whether the model is underperforming or performing as it should be, and whether it is still predicting the expected outcomes. Various metrics can be applied to test for the performance of the model, to assess whether it has satisfied the required outcomes of the business problem, needs to be adjusted or even discarded and a new model developed.
Realising the benefits
This being the case, it’s obvious that organisations can’t afford to take six or nine months to go through a process like this every time they have a business problem that needs solving. This is where the newer technologies, such as in-memory advanced analytics using big data, are adding a lot of value in being able to analyse data much faster, so that businesses can ask more and varied questions. This allows them to improve existing operations and identify opportunities to innovate.
In the past, organisations were limited by the amount of analytics they could carry out, using the hardware and systems that they had. Now, it’s about how quickly it can be done, asking the right questions and, ultimately, being able to implement the knowledge that has been gained by using analytics much faster than the competition. That’s where the real benefit comes for the organisation – identifying and acting on competitive information. But for that to happen at any kind of meaningful speed, HPA has to be embedded in the fabric of the organisation.
* André Zitzke, Data Scientist at SAS South Africa
* Follow Gadget on Twitter on @GadgetZA