Unsupervised lithology clustering from well logs, a case study in Ha Lam coalfield, Vietnam

Manual interpretation of massive well log data is time-consuming and prone to human bias. Machine Learning (ML) prediction is expected to be a robust tool to interpret lithology automatically. In this work, we apply unsupervised ML techniques such as K-means and Fuzzy C-Mean to the interpreted wells for lithology clustering. The four clustered dataset are then compared with the experts' facies interpretation to assess the clustering performance and to relabel them (removing human bias). Those labelled dataset will be fed into a supervised model for automating the facies interpretation work in the next phase of the project. The input well logs are Natural Gamma (NG) and Gamma - gamma (GG) logs from the wells in Ha Lam coalfield, Vietnam.