Mike Guilfoyle of ARC has been doing some interesting writing on predictive analytics. One of the terms he uses is the “citizen data scientist.” I like the concept, and will explain Mike’s thinking in this article.
But let me provide a little context first. It would be easy for a supply chain executive, or any business manager, to ask what is the difference between “predictive analytics” and “forecasting.” Forecasting is based on traditional mathematical and statistical models; predictive analytics uses machine learning, artificial intelligence (AI), and other advanced analytics to predict/forecast the future.
There are many places where statistical forecasting techniques work just fine, and where the use of AI does not improve the forecast at all. This is particularly true when there is not access to Big Data sets.
And existing supply chain applications do a good job of providing good analytics on a wide variety of key performance indicators – everything from on time performance to the cost of picking cases in a warehouse.
But there are also things that companies want to know, or should want to know, that existing off the shelf applications can’t answer. One example, in this environment of driver shortages, carriers need to be able to predict when drivers are about to leave the company and take preemptive actions to try and prevent their loss.
In short, there is a need for specialized predictive analytic solutions to help companies run their businesses better. But these solutions require data scientists, who are both expensive and can be picky about the kinds of projects they want to work on. Thus, the need for “citizen data scientists.”
The idea is that organizations can leverage internal skills to bring the basic data science expertise in house for advanced analytics while minimizing the burden on organizational resources. “Engineers can fill the role of citizen scientists, bringing to bear their background in math, statistics, and modeling. They can use these skills to wring more value from the analytics solutions they are using, as well as dig more deeply and broadly into data made available to them.” In supply departments, many organizations have industrial engineers.
Analytics providers are working to build out tools to support the citizen data scientist. They are creating data visualization tools, where data streams can be added, connected, and analyzed via drop and drag methods. These tools are powered by an underlying analytics engine, so it is easier for the citizen data scientist to create more “ah ha” moments using data, algorithms and models.
However, citizen data scientists can’t be relied upon to manage the end-to-end process of analytics development, training, and use of algorithms and models. They lack the advanced mathematical expertise to do so. There is still a need for data scientists, but not as many of them.
Further, whether or not citizen data scientists are used, a company working to build advanced analytic capabilities also needs subject matter experts (SMEs) to provide industry/process-specific context for what the patterns identified by the algorithms and models actually mean.
In conclusion, the following table that Mr. Guilfoyle created does a good job of summarizing the roles of data scientists, subject matter experts, and citizen data scientists.
|Data Scientist||Selects the algorithms, builds advanced data models, and trains and deploys solution. Responsible for the “math.”||Typically delivered by solution provider. However, tools for data scientists are increasingly being incorporated into sandbox environments within solutions, though many of these still require users to have advanced mathematical skills. As these data science skills become increasingly valuable within the market, a premium will be placed on solutions that enable in-house data scientists to be more productive (versus an end user company having to hire new data scientists).|
|Subject matter expert (SME)||Provides industry/process-specific context for what the patterns identified by the algorithms and models mean.||Typically provided by end user. Some solution providers also have this as part of their value proposition, though it is sometimes delivered via behind-the-scene partnerships.|
|Citizen Data Scientist||Has data science-like skills, such as statistics, but not as advanced as a data scientist. Delves into new data and improves existing models. Creates and deploys additional basic models to gain added insight.||Many solution providers are addressing this role via services. Some are also embedding tools within their solutions geared for users that have experience with statistics and modeling, such as engineers, but don’t require data science math expertise. As analytics initiatives move from pilot to become competencies within enterprise, this role will increase in importance for the end user company. SMEs will take on the roles of citizen scientists for advanced analytics as core job responsibilities, supported by an analytics toolkit.|