The SAS Data Science Blog
Advanced analytics from SAS data scientists
This guide explains how businesses can successfully implement generative AI by focusing on narrow use cases, curating data, leveraging AI agents, safeguarding sensitive information, monitoring for bias and toxicity, and ensuring model accuracy and relevancy.
Authors: Bahar Biller, Jagdishwar Mankala, and Jinxin Yi Managing spare parts inventory is a critical aspect of asset performance management, especially in industries where equipment downtime is costly. This post, based on a real-world project with a major aircraft manufacturer, explores how to optimize spare parts inventory under uncertainty. We
Enforce SAS code standards using SASjs and GitHub Actions to maintain secure, readable, and maintainable code. Automated linting blocks non-compliant code from merging into protected branches.
Accurately identifying lag structures between related time series is essential in public health forecasting, particularly during epidemics where delays between infections and hospitalizations affect planning. Using a simulated SEIR model and SAS Viya’s PROC TSSELECTLAG, distance correlation is shown to outperform Pearson correlation by correctly identifying nonlinear lag relationships—such as the true seven-day lag between new infections and hospital admissions.
As agentic AI systems evolve through protocols like MCP and A2A, traditional security practices must be adapted to address new risks such as goal misalignment and tool instruction abuse. This article explores practical threat modeling strategies, including goal alignment cascades and distinguishing between parameter-only vs. instruction-enabled tool calls.
Get an introduction to AI Copilot for the SAS Strategic Supply Chain Optimization Model, designed to democratize access to advanced modeling capabilities.