Falkonry handles the complexity of analyzing real-world time series data
Comprehensive Data Preparation
Time series data from operational historians, PLCs, sensors, and computers have idiosyncratic issues that should be addressed during analysis. Most custom data science projects spend upwards of 70% of their time on this work, which is repeated from one project to another. Falkonry automates this activity and frees up the subject matter expert to focus on the relationship of patterns to behaviors.
During the course of an analytical system’s operation, a lot of changes are required as new behaviors arise and existing ones look different. With Falkonry, you can manage the evolution of the AI applied to your time series data all while the running system continues to do its job. Then as you develop a stronger, improved AI model, it can take the place of your current operational model without the loss of any continuity.
Falkonry software updates are isolated from running operational processes so that users can opt into Falkonry’s new features and improvements selectively.
Temporal feature extraction
Time series data is often used with signal processing algorithms, such as FFT and wavelet transforms, to extract key signals that can be used to understand phenomenon from the original data. Traditional techniques require manual selection of parameters and their tuning to every data set.
With Falkonry, the selection of algorithms and parameters is automatically driven by the characteristics of the data, which reduces the need for a data scientist. This allows subject matter experts to manage the entire analysis effort through automated tools and reduces the cycle time.
Falkonry’s algorithms take advantage of commodity hardware capabilities to the fullest extent through the use of native code, in-memory computing, and distributed processing. The ability to complete learning revisions within minutes means that subject matter experts can utilize available hardware resources to their fullest extent and achieve rapid progress on their project.
Subject matter experts can easily introduce their know-how in the form of feedback and improve their AI model of the interesting behavior. Cycle time in analytics project is the main barrier to practical and scalable adoption and Falkonry’s design and implementation enables that.