Machine Learning Times
Machine Learning Times
EXCLUSIVE HIGHLIGHTS
Effective Machine Learning Needs Leadership — Not AI Hype
 Originally published in BigThink, Feb 12, 2024.  Excerpted from The...
Today’s AI Won’t Radically Transform Society, But It’s Already Reshaping Business
 Originally published in Fast Company, Jan 5, 2024. Eric...
Calculating Customer Potential with Share of Wallet
 No question about it: We, as consumers have our...
A University Curriculum Supplement to Teach a Business Framework for ML Deployment
    In 2023, as a visiting analytics professor...
SHARE THIS:

10 years ago
Socializing Predictive Analytics within Your Organization

 

With  the field of predictive analytics becoming a more mainstream business discipline, the end objective for many organizations is to operationalize this discipline in order to truly leverage the   full business impact. The notion of “operationalizing”  this  discipline is that its solutions are used and applied by all sectors of the organization. Predictive analytics within this context   is a key component of the corporate DNA or culture. But how do organizations evolve into this mindset keeping in mind that the operationalization of predictive analytics is still the exception rather than the norm.. Like any discipline to become mainstream, it needs to be socialized throughout the organization.   But this is easier said than done. Let’s see why?

One of the biggest barriers in the socialization of predictive analytics  is the inability to demonstrate  what  the results of predictive analytics projects  truly  mean to the overall organization. This poses a significant barrier towards the goal of “operationalizing “ predictive analytics within the corporation.  But what is causing this barrier?

The “business” people often have difficulty with this discipline as their  initial perception is that this discipline relies heavily on advanced analytics or statistics. Most business folks have limited knowledge of stats and have forgotten the basics of Stats 101 which they learned in 1st year business school. The data scientists or analytics practitioners, being well-versed in stats, will build tools and solutions where the solution’s benefits utilize statistical jargon  that is the familiar terrain of the data scientist.  However, the business people  cry the familiar refrain: “I don’t understand and what does this mean to the business” . The real bottom-line  impact of these solutions  is never effectively communicated  to those key business individuals. As a result, companies are unwilling to invest in  further  resources  thereby creating this catch-22  where analytics proponents  cry the familiar refrain “How can I demonstrate benefits without the necessary resources”.  This circular debate leads to frustration for c-suite executives and for analytics practitioners who in many cases will opt for jobs  in greener  pastures.  The real answer in addressing this dilemma is that the business people need to clearly understand the $ benefits of predictive analytics  while at the same time having  a solid understanding of what comprises the solution. This does not mean that the business people  need to take refresher courses in stats  but they do need to gain enough understanding such that predictive analytics is not viewed as this gigantic “Black Box”. But how do we accomplish this?

Let’s take the simple example   of a simple response model  to a marketing offer. In this situation, it is evident what the data scientist/analyst needs to do. Obviously, the creation of a solution that best targets  people most likely to respond is the desired objective   Statistical jargon is eliminated as it is the data scientist’s responsibility to understand the statistics behind the output. Yet, the data scientist needs to take this output and to then transform it to output that is meaningful to the business end user.   In the case of this simple  response model,  we can do the following:

  • Describe what the ideal responder would look like
  • Demonstrate the $ benefits of the model

Different types of reports such as EDA(Exploratory Data Analysis Reports),decile/gains tables,etc,  can be produced  with the objective of  conveying learning and insight  that the business understands . But reports in and of themselves are only one step in the socialization of predictive analytics. Verbal communication is the 2nd key component here. This can comprise  one on one meetings or small group  meetings with the business end users. These more informal type meetings, albeit requiring more work, allow  the analyst to build a network of strong support in this area.

At one organization ,we built credit-risk models where  every Friday morning, small informal meeting were held to different groups of client service reps. Both myself(the modeler) and the director of  client services conducted sessions with a short presentation along with a Q & A that would help to illustrate what the solutions meant to the client service rep. The client-service reps were then able to integrate these solutions as part of their decision-making process with clients.

Through this  above type approach , we are in effect building a network of strong allies for that all-important  C-Suite presentation  for more funding as we attempt to expand this discipline throughout the organization with the end objective being the operationalization of predictive analytics throughout the organization. But  socialization represents the “soft skill” of predictive analytics that is most required to achieve our end objective.

By: Richard Boire, Partner, Boire Filler Group
Originally published at langtechnews.hivefire.com

Leave a Reply