This work proposes an approach to estimate the clay content of shaly sand formations using Machine Learning
techniques. Currently, the main methodology used to investigate shaliness is based on a set of parameters defined
by interpreters. In this manuscript, we present an nonparametric approach to analyze well logs aiming at automatically estimating such parameters by performing two straightforward steps: i) firstly, a clustering algorithm
is used to extract patterns to confirm the gamma-ray log bimodal distribution behavior; ii) secondly, we use
Gaussian mixture models to calculate the shales and sandstones means and standard deviations to set the necessary parameters to obtain the shaliness. Another strategy, adopting a similar methodology, was perfomed using
the resistivity log to compute the shaliness. Experiments emphasize the importance of our approach, presenting
outstanding results and allowing us to identify important petrophysical properties, like the shaliness in shaly
sand models, in accordance with interpreters opinion