Event details

Corrosion is the cause for 15% of the leakage occurrences of pipelines in UK (HSE UK). Extreme distributions are often used in the corrosion data analysis, such as prediction of remaining life of buried pipelines, the maximum of coating defect area and the maximum corrosion depth. Two major challenges in analyzing massive corrosion data sets from different locations are storage and computational efficiency. This talk presents an extreme value regression method for adapting aggregation and compression technique in the massive data analysis.