Methods for automating aspects of machine learning, and related systems and apparatus
Various systems and methods provide an intuitive user interface that enables automatic specification of queries and constraints for analysis by ML component. Various implementations provide methodologies for automatically formulating machine learning (“ML”) and optimization queries. The automatic generation of ML and/or optimization queries can be configured to use examples to facilitate formulation of ML and optimization queries. One example method includes accepting input data specifying variables and data values associated with the variables. Within the input data any unspecified data records are identified, and a relationship between the variables specified in the input data and a variable associated with the at least one unspecified data record is automatically determined. The relationship can be automatically determined based on training data contained within the input data. Once a relationship is established a ML problem can be automatically generated.
1. A method, comprising:
accessing, by a computer system, time series data comprising a sequence of data records, each of the data records in the sequence corresponding to a respective time and indicating data values of a first variable and a second variable for the respective time;
generating, by the computer system via application of machine learning to the sequence of data records, a function that computes a data value of the first variable for a first time based at least in part on respective data values of at least the second variable for a plurality of times prior to the first time; and
determining, by the computer system using the function generated via application of machine learning to the sequence of data records, a data value of the first variable for a second time subsequent to one or more times corresponding to any of the data records in the sequence.
2. A system, comprising:
one or more processors connected to memory to:
access time series data comprising a sequence of data records, each of the data records in the sequence corresponding to a respective time and indicating data values of a first variable and a second variable for the respective time;
generate, via application of machine learning to the sequence of data records, a function that computes a data value of the first variable for a first time based at least in part on respective data values of at least the second variable for a plurality of times prior to the first time; and
determine, using the function generated via application of machine learning to the sequence of data records, a data value of the first variable for a second time subsequent to one or more times corresponding to any of the data records in the sequence.
3. The method of claim 1 , comprising:
determining, by the computer system, a confidence value associated with the data value of the first variable for the second time subsequent to the one or more times determined using the function.
4. The method of claim 1 , comprising:
converting, by the computer system, one or more values of the second variable in the sequence of data records into binary values.
5. The method of claim 1 , wherein the machine learning comprises symbolic regression, comprising:
training, by the computer system, a model using the machine learning; and
generating, by the computer system, the function using the model.
6. The method of claim 1 , comprising:
training, by the computer system using the machine learning, a plurality of candidate models for the function that computes the data value of the first variable for the first time and the respective data values of at least the second variable for the plurality of times prior to the first time.
7. The method of claim 1 , comprising:
determining, by the computer system using the function and based at least on part on the data value of the first variable determined for the second time, a data value of the first variable for a third time subsequent to the second time.
8. The method of claim 1 , comprising:
providing, by the computer system, the function generated via application of the machine learning for display via a graphical user interface.
9. The system of claim 2 , wherein the one or more processors are further configured to determine a confidence value associated with the data value of the first variable for the second time subsequent to the one or more times determined using the function.
10. The system of claim 2 , wherein the one or more processors are further configured to convert one or more values of the second variable in the sequence of data records into binary values.
11. The system of claim 2 , wherein the machine learning comprises symbolic regression, and the one or more processors are further configured to:
train a model using the machine learning; and
generate the function using the model.
12. The system of claim 2 , wherein the one or more processors are further configured to train, using the machine learning, a plurality of candidate models for the function that computes the data value of the first variable for the first time and the respective data values of at least the second variable for the plurality of times prior to the first time.
13. The system of claim 2 , wherein the one or more processors are further configured to determine, using the function and based at least on part on the data value of the first variable determined for the second time, a data value of the first variable for a third time subsequent to the second time.
14. The system of claim 2 , wherein the one or more processors are further configured to provide the function generated via application of the machine learning for display via a graphical user interface.
15. The method of claim 3 , comprising:
providing, by the computer system, an indication of the confidence value for display via a display device.
16. The method of claim 3 , comprising:
providing, by the computer system, a visual indication of the confidence value that indicates a higher level of confidence relative to a lower level of confidence associated with a second value.
17. The method of claim 6 , comprising:
determining, by the computer system, respective confidence levels associated with the plurality of candidate models; and
selecting, by the computer system, a candidate model from the plurality of candidate models based on the respective confidence levels to use to generate the function.
18. The system of claim 9 , wherein the one or more processors are further configured to provide an indication of the confidence value for display via a display device.
19. The system of claim 9 , wherein the one or more processors are further configured to provide a visual indication of the confidence value that indicates a higher level of confidence relative to a lower level of confidence associated with a second value.
20. The system of claim 12 , wherein the one or more processors are further configured to:
determine respective confidence levels associated with the plurality of candidate models; and
select a candidate model from the plurality of candidate models based on the respective confidence levels to use to generate the function.