Accelerate Learning and Execution
Today, modern analytic teams are trying new things—experimenting with new and combined approaches, tools, visualizations, and algorithms to uncover patterns in the growing mass of data. By trying new things, experimenting and transferring lessons from one industry and problem to a completely different industry and problem, modern analytic teams have significantly accelerated their learning and are driving new business value. However, to foster this level of innovation through experimentation, there has to be a culture that tolerates and expects failure as a path to learning and improving.
As an example, as data sizes have increased, modern analytic teams have started shifting away from constrained, statistical-only approaches to predictive and machine-learning approaches that can leverage the power of all of the data. One of the key lessons learned as data has increased is that the underlying tools and infrastructure need to minimize data movement in order to meet business objectives, especially for service-level agreements. Modern analytic teams have quickly realized the transferability of this lesson to various types of analysis and have incorporated these lessons into requirements at the onset.
Today, by industry standards, 60–80% of an analyst’s development time is spent doing data preparation or data munging. Another valuable lesson that modern analytic teams have discovered is that the upfront manual data munging should be minimized, and instead, data prep tasks should be automated and/or handled as part of the analytics processing activity. This dovetails with the need for businesses to move at a faster pace to be ahead of the competition. The ability of an organization to learn in as close to real-time as possible is a trend whose momentum will continue to build. The ability to uncover patterns in real-time, act on them almost instantaneously, and continue to discover deeper insights to improve the next cycle is a requirement in the modern business world.