Correlation-based imputation method for estimating missing well log data in a Hungarian groundwater well




missing value, data matrix, correlation, imputation


Incomplete well logging datasets measured in boreholes are frequently encountered especially in old wells. The main goal of the research, described in this paper, is to refill a data matrix with synthetic data that is missing for some reason and how the imputation procedure was to determine well the measured and estimated values fit in the case of data gaps of different sizes, and how the clay content and porosity change as a result. There can be many reasons for the lack of data, such as not measuring the desired parameters in a given depth zone. To fill this data gap, correlation-based imputation method was used which was applied in MATLAB software development system. A case study involving a Hungarian thermal water well is shown to demonstrate the reliability of the multi-linear correlation based method, which can be fruitfully applied in other wells and investigation areas.