Graphical user interface for automated machine learning model development with data health assessment
Various embodiments are generally directed to techniques for intuitive machine learning (ML) development and optimization, such as for application in a content services platform (CSP), for instance. Many embodiments include a ML model developer and a ML model evaluator to provide a graphical user interface that guides ML layman in developing, evaluating, implementing, managing, and/or optimizing ML models. Some embodiments are particularly directed to a common interface that provides a step-by-step user experience to develop and implement ML techniques. For example, embodiments may include computing a health score for various aspects of developing and/or optimizing ML models, and using the health score, and the factors contributing thereto, to guide production of a valuable ML model. These and other embodiments are described and claimed.
1 . A system comprising:
one or more processors; and
a memory comprising instructions that when executed by the one or more processors cause the one or more processors to:
present, via a graphical user interface (GUI), a plurality of data sets available for development of machine learning (ML) models, wherein each data set comprises a plurality of samples;
identify, based on input received via the GUI, a data set of the plurality of data sets for development of a ML model;
determine a plurality of characteristics associated with each sample in the data set identified, wherein a value for each of the plurality of characteristics is assignable to each sample in the data set identified;
compute a health of each characteristic in the plurality of characteristics, wherein health of a respective characteristic indicates a predictability of values for the respective characteristic;
present, via the GUI, the plurality of characteristics associated with the data set and the health of each of the plurality of characteristics;
identify, based on input received via the GUI, a target characteristic to predict values for; and
identify, based on input received via the GUI, a set of predictor characteristics to
utilize in prediction of values for the target characteristic; and
develop the ML model to predict values for the target characteristic based on values for the set of predictor characteristics.
2 . The system of claim 1 the memory comprising instructions that when executed by the one or more processors cause the one or more processors to:
determine an action to improve the health of a characteristic of the plurality of characteristics; and
present the action to improve the health of the characteristic via the GUI.
3 . The system of claim 2 , wherein the action to improve the health of the characteristic comprises one or more of adding samples to the data set identified and assigning values for the characteristic to one or more samples.
4 . The system of claim 1 , the memory comprising instructions that when executed by the one or more processors cause the one or more processors to compute the health of each characteristic in the plurality of characteristics with a ML algorithm.
5 . The system of claim 1 , the memory comprising instructions that when executed by the one or more processors cause the one or more processors to:
compute a health of each data set in the plurality of data sets, wherein health of a respective data set indicates a potential for training an accurate ML model based on the respective data set; and
present, via the GUI, the plurality of data sets available for development of machine learning (ML) models and the health of each of the plurality of data sets.
6 . The system of claim 5 , the memory comprising instructions that when executed by the one or more processors cause the one or more processors to:
determine an issue with the data set of the plurality of data sets to compute a health of the data set; and
present, via the GUI, the issue with the data set.
7 . The system of claim 1 , wherein the system is a distributed computing system, wherein different or same instructions are executed on multiple processors in communication over the Internet.
8 . The system of claim 7 , the memory comprising instructions that when executed by the one or more processors cause the one or more processors to:
determine a set of candidate characteristics based on the target characteristic;
compute a health of each candidate characteristic, wherein health of a respective candidate characteristic indicates potential to be a predictor for the target characteristic; and
present, via the GUI, the set of candidate characteristics and the health of each of the candidate characteristics.
9 . The system of claim 8 , the memory comprising instructions that when executed by the one or more processors cause the one or more processors to:
determine an action to improve the health of a candidate characteristic in the set of candidate characteristics; and
present the action to improve the health of the candidate characteristic via the GUI.
10 . The system of claim 9 , wherein the action to improve the health of the characteristic comprises assigning values for the characteristic to one or more samples.
11 . The system of claim 1 , wherein the data set comprises text-based documents and the target characteristic is associated with text in the text-based document.
12 . At least one non-transitory computer-readable medium comprising a set of instructions that, upon execution by one or more processors, cause the one or more processors to:
present, via a graphical user interface (GUI), a plurality of data sets available for development of machine learning (ML) models, wherein each data set comprises a plurality of samples;
identify, based on input received via the GUI, a data set of the plurality of data sets for development of a ML model;
determine an least one characteristic associated with the plurality of samples in the data set identified, wherein a value for each of the at least one characteristics is assignable to the plurality of samples in the data set identified;
compute a health of the at least one characteristic in the plurality of characteristics, wherein health of a respective characteristic indicates a predictability of values for the respective characteristic;
present, via the GUI, the at least one characteristic associated with the data set identified and the health of the at least one characteristic;
identify, based on input received via the GUI, a target characteristic to predict values for;
develop the ML model to predict values for the target characteristic; and
either or both:
retrain the ML model based on input received to form a new version of the ML model, and present, via the GUI, the at least one characteristic associated with the data set identified and the health of the at least one characteristic based on the new version of the ML model; or
present, via the GUI, after new content is added to the data set identified to form a new data set, at least one characteristic associated with the new data set and the health of the at least one characteristic associated with the new data set.
13 . The at least one non-transitory computer-readable medium of claim 12 , comprising instructions that, in response to being executed by the one or more processors cause the one or more processors to:
determine an issue with a characteristic of the plurality of characteristics to compute a health of the characteristic; and
present the issue with the characteristic via the GUI;
wherein the issue with the characteristic comprises one or more of insufficient samples in the data set identified and insufficient values in the samples for the characteristic.
14 . The at least one non-transitory computer-readable medium of claim 13 , comprising instructions that, in response to being executed by the one or more processors cause the one or more processors to:
determine a possible solution to the issue with the characteristic; and
present the possible solution to the issue via the GUI.
15 . The at least one non-transitory computer-readable medium of claim 12 , comprising instructions that, in response to being executed by the one or more processors cause the one or more processors to present, via the GUI, a link to add samples to the data set identified or assign values for the characteristic to one or more samples.
16 . The at least one non-transitory computer-readable medium of claim 12 , comprising instructions that, in response to being executed by the one or more processors cause the one or more processors to:
retrain the ML model based on input received to form a new version of the ML model, and present via the GUI, the at least one characteristic associated with the new data set and the health of the at least one characteristic based on the new version of the ML model.
17 . The at least one non-transitory computer-readable medium of claim 12 , comprising instructions that, in response to being executed by the one or more processors cause the one or more processors to:
present, via the GUI, after the new content is added to the data set to form a new data set, at least one characteristic associated with the new data set and the health of the at least one characteristic associated with the new data set.
18 . The at least one non-transitory computer-readable medium of claim 12 , comprising instructions that, in response to being executed by the one or more processors cause the one or more processors to:
display on the GUI, performance trends associated with the health of the at least one characteristic of the ML model or the new ML model, or both as the new content is added to the data set to form the new data set.
19 . A computer-implemented method, comprising:
presenting, via a graphical user interface (GUI), a plurality of data sets available for development of machine learning (ML) models, wherein each data set comprises a plurality of samples;
receiving input via the GUI, to identify a data set of the plurality of data sets for development of an ML model;
determining or receiving information on a determination of a plurality of characteristics associated with a plurality of samples in the data set identified, wherein a value for at least one of the plurality of characteristics is assignable to each sample in the data set identified;
receiving input via the GUI identifying a target characteristic to predict values for;
receiving a determination of or determining a set of candidate characteristics based on the target characteristic;
computing or receiving information on a computation of a health of at least one characteristic in the plurality of characteristics, wherein health of a respective characteristic indicates a predictability of values for the respective characteristic;
computing or receiving a computation of a health of at least one candidate characteristic, wherein health of a respective candidate characteristic indicates potential to be a predictor for the target characteristic; and
presenting, via the GUI, the set of candidate characteristics and the health of at least one of the candidate characteristics.
20 . The computer-implemented method of claim 19 , comprising:
presenting, via the GUI, the plurality of characteristics associated with the data set identified and the health of each of the plurality of characteristics;
developing the ML model to predict values for the target characteristic.