Probability distribution assessment for classifying subterranean formations using machine learning
According to some aspects, machine-learning models can be executed to classify a subsurface rock. Examples include training numerous machine-learning models using training data sets with different probability distributions, and then selecting a model to execute on a test data set. The selection of the model may be based on the similarity of each data point of the test data set and the probability distribution of each training class. Examples include detecting and recommending a pre-trained model to generate outputs predicting a classification, such as a lithology, of a test data set. Recommending the trained model may be based on calculated prior probabilities that measure the similarity between the training and test data sets. The model with a training data set that is most similar to the test data set can be recommended for classifying a physical property of the subsurface rock for hydrocarbon formation.
1. A system comprising:
a processor; and
a non-transitory memory device comprising instructions that are executable by the processor to cause the processor to perform operations comprising:
receiving geological or geophysical data collected for hydrocarbon exploration from a plurality of subterranean formations;
generating, using the geological or geophysical data, a plurality of training data sets used to train machine-learning models, each training data set of the plurality of training data sets including a portion of the geological or geophysical data and one or more training classes, each training class being defined by a probability distribution and representing a classification of the portion of the geological or geophysical data;
receiving a test data set including geological or geophysical data of a subterranean formation;
selecting a training data set from the plurality of training data sets by comparing the test data set with the probability distribution of each training class of the one or more training classes of the training data set;
determining a trained machine-learning model trained using the selected training data set;
generating an output representing a recommendation to execute the trained machine-learning model to classify the test data set; and
displaying, based on a result of executing the trained machine-learning model on the test data set, an interface presenting a classification of the test data set that is usable for hydrocarbon exploration.
2. The system of claim 1 , wherein the operations further comprise:
inputting the test data set into the trained machine-learning model associated with the selected training data set;
generating an additional output, using the trained machine-learning model, the additional output predicting a characteristic of the subterranean formation; and
generating, based on the additional output, a plan for drilling a wellbore for hydrocarbon exploration.
3. The system of claim 1 , wherein the operation of selecting further comprises:
determining, for each data point of a plurality of data points of the test data set, a distance between the data point and the probability distribution of each training class of the one or more training classes; and
identifying the training class associated with a smallest distance between the data point and the probability distribution of the training class as compared to distances of remaining data points of the plurality of data points of the test data set.
4. The system of claim 3 , wherein the operation of determining the distance further comprises:
determining a Mahalanobis distance.
5. The system of claim 1 , wherein the operation of selecting further comprises:
determining a similarity metric representing a similarity between a data point of the test data set and the probability distribution of a training class of the one or more training classes of a training data set; and
selecting, based on the determined similarity metric, the trained machine-learning model for classifying the test data set.
6. The system of claim 1 , wherein the operation of generating the plurality of training data sets further comprises:
defining each training class of the one or more training classes of a training data set by labeling a subset of the training data set with the classification, wherein the classification is a lithology.
7. The system of claim 1 , wherein the operations further comprise:
generating a chi-squared distribution of a distance between a data point of the test data set and the probability distribution of each training class; and
selecting the training data set from the plurality of training data sets based on the chi-squared distribution.
8. A computer-implemented method comprising:
receiving geological or geophysical data collected for hydrocarbon exploration from a plurality of subterranean formations;
generating, using the geological or geophysical data, a plurality of training data sets, each training data set of the plurality of training data sets including a portion of the geological or geophysical data and one or more training classes, each training class being defined by a probability distribution and representing a classification of the portion of the geological or geophysical data;
training a plurality of machine-learning models based on the plurality of training data sets, each machine-learning model of the plurality of machine-learning models being trained using a respective training data set of the plurality of training data sets; and
subsequent to training the plurality of machine-learning models:
receiving a test data set including geological or geophysical data of a subterranean formation;
selecting a training data set from the plurality of training data sets, by comparing the test data set with at least one respective probability distribution associated with each respective training data set of the plurality of training data sets;
selecting, from among the plurality of machine-learning models, a trained machine-learning model that was trained using the selected training data set;
generating an output representing a recommendation to execute the trained machine-learning model to classify the test data set; and
displaying, based on a result of executing the trained machine-learning model on the test data set, an interface presenting a classification of the test data set that is usable for hydrocarbon exploration.
9. The computer-implemented method of claim 8 , further comprising:
inputting the test data set into the trained machine-learning model associated with the selected training data set;
generating an additional output, using the trained machine-learning model, the additional output predicting a characteristic of the subterranean formation; and
generating, based on the additional output, a plan for drilling a wellbore for hydrocarbon exploration.
10. The computer-implemented method of claim 8 , wherein the selecting further comprises:
determining, for each data point of a plurality of data points of the test data set, a distance between the data point and the probability distribution of each training class of the one or more training classes; and
identifying the training class associated with a smallest distance between the data point and the probability distribution of the training class as compared to distances of remaining data points of the plurality of data points of the test data set.
11. The computer-implemented method of claim 10 , wherein determining the distance further comprises:
determining a Mahalanobis distance.
12. The computer-implemented method of claim 8 , wherein the selecting further comprises:
determining a similarity metric representing a similarity between a data point of the test data set and the probability distribution of a training class of the one or more training classes of a training data set; and
selecting, based on the determined similarity metric, the trained machine-learning model for classifying the test data set.
13. The computer-implemented method of claim 8 , further comprising:
defining each training class of the one or more training classes of a training data set by labeling a subset of the training data set with the classification, wherein the classification is a lithology.
14. The computer-implemented method of claim 8 , further comprising:
generating a chi-squared distribution of a distance between a data point of the test data set and the probability distribution of each training class; and
selecting the training data set from the plurality of training data sets based on the chi-squared distribution.
15. A computer-program product tangibly embodied in a non-transitory machine-readable storage medium, including instructions configured to cause a processing apparatus to perform operations including:
receiving geological or geophysical data collected for hydrocarbon exploration from a plurality of subterranean formations;
generating, using the geological or geophysical data, a plurality of training data sets, each training data set of the plurality of training data sets including a portion of the geological or geophysical data and one or more training classes, each training class being defined by a probability distribution and representing a classification of the portion of the geological or geophysical data;
training a plurality of machine-learning models based on the plurality of training data sets, each machine-learning model of the plurality of machine-learning models being trained using a respective training data set of the plurality of training data sets; and
subsequent to training the plurality of machine-learning models:
receiving a test data set including geological or geophysical data of a subterranean formation;
selecting a training data set from the plurality of training data sets, by comparing the test data set with at least one respective probability distribution associated with each respective training data set of the plurality of training data sets;
selecting, from among the plurality of machine-learning models, a trained machine-learning model that was trained using the selected training data set;
generating an output representing a recommendation to execute the trained machine-learning model to classify the test data set; and
displaying, based on a result of executing the trained machine-learning model on the test data set, an interface presenting a classification of the test data set that is usable for hydrocarbon exploration.
16. The non-transitory machine-readable storage medium of claim 15 , wherein the operations further comprise:
inputting the test data set into the trained machine-learning model associated with the selected training data set;
generating an additional output, using the trained machine-learning model, the additional output predicting a characteristic of the subterranean formation; and
generating, based on the additional output, a plan for drilling a wellbore for hydrocarbon exploration.
17. The non-transitory machine-readable storage medium of claim 15 , wherein the operation of selecting further comprises:
determining, for each data point of a plurality of data points of the test data set, a distance between the data point and the probability distribution of each training class of the one or more training classes; and
identifying the training class associated with a smallest distance between the data point and the probability distribution of the training class as compared to distances of remaining data points of the plurality of data points of the test data set.
18. The non-transitory machine-readable storage medium of claim 17 , wherein the operation of determining the distance further comprises:
determining a Mahalanobis distance.
19. The non-transitory machine-readable storage medium of claim 15 , wherein the operation of selecting further comprises:
determining a similarity metric representing a similarity between a data point of the test data set and the probability distribution of a training class of the one or more training classes of a training data set; and
selecting, based on the determined similarity metric, the trained machine-learning model for classifying the test data set.
20. The non-transitory machine-readable storage medium of claim 15 , wherein the operation of generating further comprises:
defining each training class of the one or more training classes of a training data set by labeling a subset of the training data set with the classification, wherein the classification is a lithology.