Explore

Warning

Demonstrating Predictive and Prescriptive Analytics can be BORING.

Bored

Numbers in, crunch, crunch, and number out.

  1. Historic Data is gathered periodically
  2. Data is processed from single or multiple sources
  3. The process often uses multiple formulae
  4. Predictive Data is output into spreadsheets, files, charts, dashboards
  5. Prescriptions are sometimes used to launch action items like email or alerts
  6. Continual improvements are made and the process is repeated

This presentation is designed to introduce you to several common models that our customers appreciate and depend upon.  Note: We have dozens of models for you to choose from when you become a customer!

Let’s begin a brief introduction to Predictive & Prescriptive Analytics.
Aka: Machine Learning and Artificial Intelligence.

Artificial Intelligence in the form of Machine Learning is not a new concept

  • For decades companies have asked their software to make purchasing decisions for them.
  • Another example is/was getting a system-generated warning when a customer is becoming past due.

What changed

Many more decisions can be made now than before!  We can predict things you may not have known were possible.  We can help you reduce risk, save and make more money.

Please Take a Moment to Consider The Following

Ponder

A great risk to predictions begins with the belief that they need to be perfect or exact.  Being close is better than being totally caught off guard. If the system detects that a customer is going to leave us, it doesn’t need to tell us on which day.  It doesn’t matter much if they leave in a month or three. They’d be missed and we have tools to give us time to save them!

In the past, secondary files or outside influences made predictive analysis much more complicated. Using multiple data sources is now feasible and a standard practice in providing many of our predictions.

The massive volume of numbers needed to be crunched was, and in some cases still is, beyond companies internal system’s capabilities.  We can now provide this horsepower without breaking the bank or forcing you to replace your internal systems.

Predictive and Prescriptive Analytics are now attainable

Please select the example you would like to review.

Example 1 Cross-Selling – they’re about to buy Prodx, why not suggest Prodt and Prodv?

Example 2 Trending – Looking for extremes in all aspects of the business.

Example 3 Customer Churn – Predict when a customer shows signs of leaving.

Example 4 Runaway Shipping – helping control runaway shipping costs.

Machine Learning is an art and a science. 

CMi uses appropriate algorithms

  • Neural Net and Decision Forest algorithms have both performed well in different situations for Customer Churn
  • Decision Forest and specialized regression algorithms do well for trending over transaction data

Some model processes need more customization than others

  • Either
    • fit models to business processes
    • Or, fit business processes to models

Return to How

 

 

 

 

%d bloggers like this: