Data preparation in machine learning, or seismic inversion and in subsurface exploration in general is paramount. While we can arbitrarily modify the input, bias could be including unintentionally, disserving the data conditioning purpose. The objective of the lecture is to showcase the use of statistical methodologies to prepare the data for seismic inversion, both seismic information and borehole data. Additionally, rock-physics models are to be assist the borehole data preparation. Basic concepts of rock-physics and geostatistics are going to be reviewed in depth as well as seismic data conditioning for partial stacks, with a particular focus in amplitude preservation and enhancing fault visibility. The relevance and effect of data conditioning is going to be displayed on a seismic simultaneous elastic inversion. Additionally, the lecture will show shallowly how to interpret the elastic attributes as a function of petrophysical parameters using the Rock-physics template, and the two generated datasets (conditioned and non-conditioned) will be used to evaluate the differences.