The New York Section (New Jersey) of the Society for Applied Spectroscopy (SAS) is the first and oldest spectroscopy networking organization in the United States. We were formed during World War II, before the parent organization of SAS. Our first Gold Medal Award symposium was awarded to Dr. William F. Meggers in 1952. The local section provides a forum for practicing spectroscopists to interact, lead, and share insights that can benefit their respective organizations. Monthly meetings feature research presentations from academia, government, and industry, and provide members with opportunities to expand their knowledge and application of vibration, electronic, and atomic spectroscopy. For more information, click on the "About NYSAS" page.
NY/NJ Monthly Meetings are FREE Zoom Meetings open to all International SAS Members and interested Scientists
Invited Seminar – Thursday, August 20, 2026, 12:00 – 1:00 PM EDT
KNNR in the Wild: Nonlinear Neighborhoods that Outperform PLS (Sometimes)
Speaker: Dr. David Honigs │ PerkinElmer, Perten Division
Join Zoom Meeting (See Meeting Details Below)
https://us02web.zoom.us/j/89243994684?pwd=YJzX44JdegRJHYPyIvtxJ9xMEHqsF8.1
ABSTRACT
K-Nearest Neighbor Regression (KNNR) isn’t new—but applying it to high-dimensional NIR spectral data in real-world conditions reveals surprising strengths. In this work, we examine cases where KNNR outperforms Partial Least Squares (PLS), not by modeling global latent structure, but by embracing local, resilient patterns that align more closely with how samples vary in practice.
Unlike PLS, which attempts to project the entire dataset into a compressed, linear latent space, KNNR responds to structure in immediate spectral neighborhoods. This approach proves effective when the sample space is irregular, noisy, or distorted by calibration transfer, ingredient variability, or instrument drift. In such cases, global structure may be misleading—while local relationships retain predictive power.
We explore examples across typical applied environments—ranging from food production to agricultural materials—where KNNR holds its own or surpasses PLS in predictive accuracy, stability, or adaptability. Rather than proposing KNNR as a universal replacement, we highlight it as a practical and complementary alternative that thrives under specific conditions where PLS begins to stretch or overgeneralize.
What emerges is a model selection story grounded in behavioral observation rather than theoretical purity. When KNNR and PLS disagree, it often points to something deeper: nonlinear variation, clustered domains, or edge cases where local modeling sees what global methods blur. The differences aren’t noise—they are signals worth examining.
This presentation offers a technical overview of KNNR behavior in spectral applications, key comparative performance results, and guidance on when to include KNNR in the calibration stack. Our findings suggest that KNNR, often overlooked in chemometric circles, deserves renewed attention—not as a throwback, but as a tool perfectly suited for spectral data that doesn't play by linear rules.
BRIEF BIOGRAPHY
Dr. David Honigs is a PerkinElmer Field Applications Scientist specializing in Food. Dr. Honigs did his graduate work under joint supervision of Professor Gary Hieftje and Dr. Tomas Hirschfeld at Indiana University in Bloomington. He served as an Assistant Professor of Analytical Chemistry at the University of Washington for a few years. Following that he worked at NIRSystems (now part of FOSS) on NIR instruments. He started a company, Katrina Inc. which made process NIR instruments. For the last almost 25 years he has worked at Perten (now a part of PerkinElmer) on Near Infrared Instrumentation and applications in the food industry. He has 35 research papers as listed on ResearchGate.com. Dr. Honigs has 10 issued US patents. He is known for the Honigs Regression which is a non-linear approach melding KNN and PLS methods.
Zoom Meeting Details
Topic: Dr. David Honigs - KNNR in the Wild: Nonlinear Neighborhoods that Outperform PLS (Sometimes)
Time: Aug 20, 2026 12:00 PM Eastern Time (US and Canada)
Join Zoom Meeting:
https://us02web.zoom.us/j/89243994684?pwd=YJzX44JdegRJHYPyIvtxJ9xMEHqsF8.1
Meeting chat link
https://us02web.zoom.us/launch/jc/89243994684
Meeting ID: 892 4399 4684
Passcode: 794458
One tap mobile:
+16469313860,,89243994684#,,,,*794458# US
+19292056099,,89243994684#,,,,*794458# US (New York)
Join Instructions:
https://us02web.zoom.us/meetings/89243994684/invitations?signature=Y5N0wVV471uzOZ2Mf-g2hGs1Ok2y-4GwPMp0EMfaDyE
Proud sponsors of NY/NJ SAS - We thank you for your support.
