Anomaly detection algorithms do exactly what the name suggests – detect anomalies in the data. The simplest example is a sensor monitoring your body temperature, and leveraging anomaly detection. If you had a sensor measuring your body temperature, and the date is being fed to an anomaly detector in the form of time series data (like your body temperature every one hour), if the temperature value exceeds 99 degrees Fahrenheit, the anomaly detector will flag the temperature as possible mild fever. Obviously, this example is very simple but the gist of these algorithms is the same. These algorithms look at data to identify the data points that are inconsistent with the rest of the data and are an outlier or anomaly. Anomaly detection algorithms can be applied to a variety of scenarios, including credit card fraud, cyber security and asset management and maintenance. These are the areas it is already actively being used in, but the opportunities to leverage these algorithms are plenty. This is the reason that all leading hyperscaler platforms provide anomaly detection features. These may differ in the choice of machine learning algorithm ensembles though. For example, Azure cognitive services leverages SR-CNN among other algorithms, AWS SageMaker leverages random cut forest, Google Cloud Platform and SAP HANA Predictive Analytics Library (PAL) leverage k-means clustering, among other algorithms.

Anomaly detection algorithm in SAP HANA Predictive Analytics Library (PAL)

SAP HANA Predictive Analytics Library (PAL) is an Application Function Library (AFL) defines functions that can be called from within SAP HANA SQLScript procedures to perform analytic algorithms. For those interested in a deep dive, the comprehensive documentation is here. The library provides a rich portfolio  of predictive analytics algorithms in nine categories, which includes clustering algorithms for anomaly detection. The SAP HANA Smart Data Streaming capability allows you to leverage these algorithms for building dynamic models. In the world of supply chains, a key role anomaly detection algorithms play is in manufacturing, specifically in equipment management and maintenance (preventive and predictive). However, the opportunities to leverage these algorithms extend beyond the conventional. For SAP users specifically, having a library of algorithms available within SAP HANA allows seamless integration of algorithms with your data. There are some areas in supply chain where there is potential to leverage anomaly detection algorithms. An example is inventory classification. Often, materials may get categorized arbitrarily or randomly. Using anomaly detection, you can identify these misclassifications. Since safety stock calculations are typically based on these classes, this exercise can help you reduce your inventory levels as well. Inventory management and optimization is a key focus area for SAPinsiders, as highlighted in our recent research report, Inventory Management and Optimization. Identifying and leveraging these capabilities within SAP technology ecosystem will help create competitive differentiators.