Paper 90

Introduction to developing Neural Nets

September 23, 2025: 10:30 AM - 11:00 AM

Statistics, Modeling, and Analytics (SMA), Executive Dining Room #1

Authors	Abstract
David Horvath	Neural nets have been discussed in in SAS regional and international conferences since last century, This session will be a bit different in that it starts with the obligatory basic background and then goes into actual code. We will discuss what Machine Learning can and cannot solve. There will be more discussion of the how results are produced and less about the overall concepts. Topics * What is a neuron? * How does a neural net model a human neuron? * How does that work with data? * What does it NOT solve? Discuss the biggest problems with data. * Examples 1. Working data 2. Why’d we get that result? 3. More Working data 4. Why’d we get that result? 5. Reviewing the code

David Horvath

Neural nets have been discussed in in SAS regional and international conferences since last century, This session will be a bit different in that it starts with the obligatory basic background and then goes into actual code. We will discuss what Machine Learning can and cannot solve. There will be more discussion of the how results are produced and less about the overall concepts. Topics * What is a neuron? * How does a neural net model a human neuron? * How does that work with data? * What does it NOT solve? Discuss the biggest problems with data. * Examples 1. Working data 2. Why’d we get that result? 3. More Working data 4. Why’d we get *that* result? 5. Reviewing the code