Lithology prediction is considered an essential requirement in the field of petroleum exploration. Since reservoirs consist of complex lithologies, predicting the lithology classes is gradually playing a pivotal role in the geosciences. During drilling operations the advancements of real time data recording have been so common in the petroleum industries in the past and majority of the logging data are recorded in real time process. However, sometimes the system encounters data loss or missing values while going through the logging procedures. Hence, the application of missing data estimation in automated lithology prediction is so essential. In this research a unique module is developed for classifying lithology from borehole log data consisting of incomplete log values by employing non-deterministic information systems apriori (NIS Apriori) algorithm. The unique characteristics of the proposed module are also presented in the paper. The research proposes certain and possible rules based on real data science semantics following the framework of NISs. By using the NIS Apriori algorithm it is proved that each rule � is determined by analyzing only a pair of �-dependent possible tables although each particular rule � is a dependant on so many possible tables. However, one of the applications of the NIS Apriori algorithm is its prospect of the handling missing values. This research proposes a white-box novel architecture to deal with the well log missing values by using the NIS Apriori algorithm which provides the results in terms of rules to classify complex lithology efficiently. © 2020, ICIC International.