Integrating geoscience data to predict formation properties

A computing system comprising:

• one or more processors;
• and
• a memory system comprising one or more non-transitory computer-readable media storing instructions that, when executed by at least one of the one or more processors, cause the computing system to perform operations, the operations comprising:
• receiving well log data for a plurality of wells;
• generating a flag based at least partially on the well log data, wherein the flag comprises a binary value;
• sorting the wells into groups based at least partially on the well log data, the flag, or both;
• building one or more models for the wells based at least partially on the well log data, the flag, and the groups, building the one or more models includes:
• generating a first footprint defining a decision boundary for a first well of the plurality of wells using a support vector machine model, at least some of the well log data associated with the first well is outside of the first footprint and at least some of the well log data associated with the first well is within the first footprint;
• comparing the first footprint to a second footprint generated using the model representing one or more second wells, the model is configured to predict one or more formation parameters based at least in part on well log data;
• determining that the first well is similar to the one or more second wells based on the comparing;
• and
• updating the model representing the one or more second wells based at least partially on the well log data from the first well, in response to determining that the first well is similar to the one or more second wells;
• predicting a formation property based at least partially upon the well log data and the one or more models;
• determining an uncertainty of the formation property using one or more prediction intervals from a quantile regression forest algorithm;
• and
• identifying a physical action to be performed in response to the formation property, the uncertainty of the formation property, or both.

A non-transitory computer-readable medium storing instructions that, when executed by at least one processor of a computing system, cause the computing system to perform operations, the operations comprising:

• receiving well log data for a plurality of wells, wherein the well log data is selected from the group consisting of gamma ray measurements, density measurements, neutron logs, and core data;
• generating a flag based at least partially on the well log data, wherein the flag comprises a binary value that identifies whether a sample in the well log data is an outlier or an inlier, wherein the flag is generated using a one-class unsupervised support vector machine model, and wherein the one-class unsupervised support vector machine model is generated on a well-by-well basis, a global basis, or both using an outlier fraction that is user-supplied or automated;
• sorting the wells into groups based at least partially on the well log data, the flag, or both;
• building a model for each of the wells based at least partially on the well log data, the flag, and the groups;
• adding the models to a library after the models are built;
• updating the models in the library when additional ground truth becomes available updating the models includes:
• generating a first footprint defining a decision boundary for a first well of the plurality of wells using a support vector machine model, at least some of the well log data associated with the first well is outside of the first footprint and at least some of the well log data associated with the first well is within the first footprint;
• comparing the first footprint to a second footprint generated using the model representing one or more second wells, the model is configured to predict one or more formation parameters based at least in part on well log data;
• determining that the first well is similar to the one or more second wells based on the comparing;
• and
• updating the model based at least partially on the well log data from the first well, in response to determining that the first well is similar to the one or more second wells;
• predicting a formation property based at least partially upon the well log data and the models, wherein the formation property is selected from the group consisting of water saturation, porosity, permeability, and compressional slowness;
• determining an uncertainty of the formation property using one or more prediction intervals from a quantile regression forest algorithm;
• and
• identifying a physical action to be performed in response to the formation property, the uncertainty of the formation property, or both, wherein the physical action is selected from the group consisting of selecting an additional well to be interpreted manually, changing a trajectory of one of the wells, and varying a property of a fluid being pumped into one of the wells.