A simple post of gratitude to our customers and to the industry.
While some productivity drivers in a tight reservoir are observable and easily measured, others require multi-variate regressions and physics-driven feature engineering.
The shale inventory wall is a real challenge for our industry and has made the news circuit this past week.
Machine Learning (ML) models are critical tools in understanding productivity drivers and predicting DSU performance, but including too many features in your ML models reduces their impact on explain-ability and generalize-ability.
Kyle LaMotta, VP of Analytics explains, “The partial dependency plots (PDP) are one of the model diagnostics that we’ll use in every DSU Design Service analysis. The PDPs are a way to visualize how a feature is contributing to a prediction.
Feature exploration is a critical part of the Petro.ai DSU Design Service (DSUDS) model development. Recommendation reports are presented in live, interactive meetings where the client’s asset team and the Petro.ai team discuss the features that need to be included and interpreted in the model.