OCCF - Leveraging Lagged Correlation Analysis for Enhanced Insights in Continuous Industrial Data
Sabrina Luftensteiner, Kilian Krikova, Roman Josef Rainer · Procedia Computer Science · 2026
In the context of continuous data, correlation analysis combined with time lags provides valuable insights into time-dependent relationships between variables. Continuous data often exhibit delayed effects, where one variable’s influence on another unfolds over time. By incorporating lags into correlation analysis, delayed correlations can be identified, revealing complex dynamics. This is particularly useful in industrial environments, where continuous processes contain parameters that respond to changes after a certain period. Lagged correlation analysis enhances the accuracy of predictive models by accounting for delayed effects, leading to improved forecasts and a deeper understanding of temporal dependencies. It can suggest potential causality, as repeated time-lagged correlations between variables may indicate underlying causal mechanisms. It has diverse applications in industrial environments, including process control, quality control, and anomaly detection. It can help identify optimal time-lagged relationships between process variables, relationships between process variables and product quality, and detect changes in underlying process dynamics. Our novel approach identifies ideal lags within correlation analysis of continuous time series data based on time-windows. By analyzing correlations between variables at various lags, the optimal lag can be found, essential for identifying time-dependent patterns and building accurate predictive models. Our experiments show that up to 98% of lags are identified correctly using simulated and real-world data, improving also the accuracy of predictive models. Overall, our lagged correlation analysis approach is a powerful method for uncovering time-dependent relationships in continuous data, leading to improved predictive modeling, decision-making, and process optimization. It enables industries to gain a deeper understanding of complex dynamics underlying their processes, driving efficiency, reliability, and cost savings.