Hierarchical tournament-based machine learning predictions
Systems and techniques for hierarchical tournament-based machine learning predictions are described herein. A machine learning selection model may be trained with training data. A configuration may be received that includes the metric and a target prediction. A set of evaluation component combinations may be selected using the machine learning selection model. Each evaluation component combination of the set of evaluation component combinations may include an algorithm, a hierarchical learning model corresponding to a level of a hierarchy, and a prediction model for the target prediction. The set of evaluation component combinations may be transmitted to a cluster of computing nodes. Output results may be received for the set of evaluation component combinations. The output results may be evaluated using the metric to determine a winning evaluation component combination. The winning evaluation component combination may be stored in storage for use in calculating future predictions for the target prediction.
1. A system for hierarchical tournament-based machine learning predictions comprising:
at least one processor: and
memory including instructions that, when executed by the at least one processor, cause the at least one processor to perform operations to:
train a machine learning selection model with training data, the machine learning selection model trained to calculate a probability that an algorithm and a model will provide output to obtain a metric;
receive a configuration that includes the metric and a target prediction;
select a set of evaluation component combinations using the machine learning selection model, wherein each evaluation component combination of the set of evaluation component combinations includes an algorithm, a hierarchical learning model corresponding to a level of a hierarchy, and a prediction model for the target prediction;
transmit the set of evaluation component combinations to a cluster of computing nodes;
receive output results for the set of evaluation component combinations;
evaluate the output results using the metric to determine a winning evaluation component combination; and
store the winning evaluation component combination in storage for use in calculating future predictions for the target prediction.
2. The system of claim 1 , wherein the machine learning selection model selects the algorithm from a set of algorithms, wherein each algorithm of the set of algorithms include logic for performing prediction calculations using the hierarchical learning model and the prediction model.
3. The system of claim 1 , wherein the hierarchical learning model is selected from a graph cube network, and wherein the hierarchy is at least a portion of the graph cube network.
4. The system of claim 1 , wherein the prediction model is selected from a graph cube network, and wherein the hierarchy is at least a portion of the graph cube network.
5. The system of claim 1 , the instructions to evaluate the output results using the metric to determine a winning evaluation component combination further including instructions that, when executed by the at least one processor, cause the at least one processor to perform operations to:
compare the output results to a baseline value for the metric; and
determine the winning evaluation component combination based on an output result of the winning evaluation component combination having the largest positive deviation from the baseline value.
6. The system of claim 1 , the instructions to evaluate the output results using the metric to determine a winning evaluation component combination further including instructions that, when executed by the at least one processor, cause the at least one processor to perform operations to:
evaluate the output results using a machine learning algorithm to determine a fit between the metric and each output result of the output results; and
determine the winning evaluation component combination based on an output result of the winning evaluation component having the closest fit with the metric.
7. The system of claim 1 , wherein the metric is an error rate.
8. At least one non-transitory machine-readable medium including instructions for hierarchical tournament-based machine learning predictions that, when executed by at least one processor, cause the at least one processor to perform operations to:
train a machine learning selection model with training data, the machine learning selection model trained to calculate a probability that an algorithm and a model will provide output to obtain a metric;
receive a configuration that includes the metric and a target prediction;
select a set of evaluation component combinations using the machine learning selection model, wherein each evaluation component combination of the set of evaluation component combinations includes an algorithm, a hierarchical learning model corresponding to a level of a hierarchy, and a prediction model for the target prediction;
transmit the set of evaluation component combinations to a cluster of computing nodes;
receive output results for the set of evaluation component combinations;
evaluate the output results using the metric to determine a winning evaluation component combination; and
store the winning evaluation component combination in storage for use in calculating future predictions for the target prediction.
9. The at least one non-transitory machine-readable medium of claim 8 , wherein the machine learning selection model selects the algorithm from a set of algorithms, wherein each algorithm of the set of algorithms include logic for performing prediction calculations using the hierarchical learning model and the prediction model.
10. The at least one non-transitory machine-readable medium of claim 8 , wherein the hierarchical learning model is selected from a graph cube network, and wherein the hierarchy is at least a portion of the graph cube network.
11. The at least one non-transitory machine-readable medium of claim 8 , wherein the prediction model is selected from a graph cube network, and wherein the hierarchy is at least a portion of the graph cube network.
12. The at least one non-transitory machine-readable medium of claim 8 , the instructions to evaluate the output results using the metric to determine a winning evaluation component combination further including instructions that, when executed by the at least one processor, cause the at least one processor to perform operations to:
compare the output results to a baseline value for the metric; and
determine the winning evaluation component combination based on an output result of the winning evaluation component combination having the largest positive deviation from the baseline value.
13. The at least one non-transitory machine-readable medium of claim 8 , the instructions to evaluate the output results using the metric to determine a winning evaluation component combination further including instructions that, when executed by the at least one processor, cause the at least one processor to perform operations to:
evaluate the output results using a machine learning algorithm to determine a fit between the metric and each output result of the output results; and
determine the winning evaluation component combination based on an output result of the winning evaluation component having the closest fit with the metric.
14. The at least one non-transitory machine-readable medium of claim 8 , wherein the metric is an error rate.
15. A method for hierarchical tournament-based machine learning predictions comprising:
training a machine learning selection model with training data, the machine learning selection model trained to calculate a probability that an algorithm and a model will provide output to obtain a metric;
receiving a configuration that includes the metric and a target prediction;
selecting a set of evaluation component combinations using the machine learning selection model, wherein each evaluation component combination of the set of evaluation component combinations includes an algorithm, a hierarchical learning model corresponding to a level of a hierarchy, and a prediction model for the target prediction;
transmitting the set of evaluation component combinations to a cluster of computing nodes;
receiving output results for the set of evaluation component combinations;
evaluating the output results using the metric to determine a winning evaluation component combination; and
storing the winning evaluation component combination in storage for use in calculating future predictions for the target prediction.
16. The method of claim 15 , wherein the machine learning selection model selects the algorithm from a set of algorithms, wherein each algorithm of the set of algorithms include logic for performing prediction calculations using the hierarchical learning model and the prediction model.
17. The method of claim 15 , wherein the hierarchical learning model is selected from a graph cube network, and wherein the hierarchy is at least a portion of the graph cube network.
18. The method of claim 15 , wherein the prediction model is selected from a graph cube network, and wherein the hierarchy is at least a portion of the graph cube network.
19. The method of claim 15 , wherein evaluating the output results using the metric to determine a winning evaluation component combination further comprises:
comparing the output results to a baseline value for the metric; and
determining the winning evaluation component combination based on an output result of the winning evaluation component combination having the largest positive deviation from the baseline value.
20. The method of claim 15 , wherein evaluating the output results using the metric to determine a winning evaluation component combination further comprises:
evaluating the output results using a machine learning algorithm to determine a fit between the metric and each output result of the output results; and
determining the winning evaluation component combination based on an output result of the winning evaluation component having the closest fit with the metric.
21. The method of claim 15 , wherein the metric is an error rate.