Customized predictive analytical model training
Methods, systems, and apparatus, including computer programs encoded on one or more computer storage devices, for training predictive models. Multiple training data records are received that each include an input data portion and an output data portion. A training data type is determined that corresponds to the training data. For example, a training data type can be determined by inputting the output data portions into one or more trained predictive classifiers. In other example, the training data type can be determined by comparison of the output data portions to data formats. Based on the determined training data type, a set of training functions are identified that are compatible with the training data of the determined training data type. The training data and the identified set of training functions are used to train multiple predictive models.
1. A method comprising:
sending, from a client device and to a predictive modeling server system, training data records, wherein:
each training data record includes an input data portion and an output data portion and the training data records are used to train one of a plurality of predictive models that are each of a different predictive model type;
the training data records do not specify a predictive model type to be trained on the training data records; and
the predictive modeling server is configured to compare the output data portions of the training data records to a plurality of data formats, and based on the comparison, determining a training data type and a corresponding predictive model type to select a predictive model to train on the training data records;
receiving, at the client device and from the predictive modeling server system, access to a selected predictive model that has been trained on the training data records and that executes on the predictive modeling server;
sending, from the client device to the selected predicted model executing on the predictive modeling server, input data records and a prediction request; and
receiving, from the selected predictive model, predictive output generated by applying the received input data records to the selected predictive model.
2. The method of claim 1 , wherein the training data records are sent through an HTTP web service.
3. The method of claim 1 , wherein receiving access to the selected predictive model providing comprises receiving a URL address for accessing the selected predictive model.
4. The method of claim 1 , wherein receiving access to the selected predictive model providing comprises receiving an interface configured to interface a trained model repository with a network communicatively coupled to the client device.
5. The method of claim 4 , wherein the interface is further configured to interface a training data repository with the network communicatively coupled to the client device.
6. The method of claim 1 , wherein the selected predictive model is selected at the predictive modeling server by:
generating a score for each of the plurality of trained predictive models, where each score represents an estimation of the effectiveness of the respective trained predictive model trained on the training data records; and
selecting a first trained predictive model from among the plurality of trained predictive models based on the generated scores.
7. A system comprising:
a client computer device configured to perform operations comprising:
sending, from the client device and to a predictive modeling server system, training data records, wherein:
each training data record includes an input data portion and an output data portion and the training data records are used to train one of a plurality of predictive models that are each of a different predictive model type;
the training data records do not specify a predictive model type to be trained on the training data records; and
the predictive modeling server is configured to compare the output data portions of the training data records to a plurality of data formats, and based on the comparison, determining a training data type and a corresponding predictive model type to select a predictive model to train on the training data records;
receiving, at the client device and from the predictive modeling server system, access to a selected predictive model that has been trained on the training data records and that executes on the predictive modeling server;
sending, from the client device to the selected predicted model executing on the predictive modeling server, input data records and a prediction request; and
receiving, from the selected predictive model, predictive output generated by applying the received input data records to the selected predictive model.
8. The system of claim 7 , wherein the training data records are sent through an HTTP web service.
9. The system of claim 7 , wherein receiving access to the selected predictive model providing comprises receiving a URL address for accessing the selected predictive model.
10. The system of claim 7 , wherein receiving access to the selected predictive model providing comprises receiving an interface configured to interface a trained model repository with a network communicatively coupled to the client device.
11. The system of claim 10 , wherein the interface is further configured to interface a training data repository with the network communicatively coupled to the client device.
12. The system of claim 7 , wherein the selected predictive model is selected at the predictive modeling server by:
generating a score for each of the plurality of trained predictive models, where each score represents an estimation of the effectiveness of the respective trained predictive model trained on the training data records; and
selecting a first trained predictive model from among the plurality of trained predictive models based on the generated scores.
13. A non-transitory computer-readable medium comprising instructions that, when executed by a client device, cause the client device to perform operations comprising:
sending, from the client device and to a predictive modeling server system, training data records, wherein:
each training data record includes an input data portion and an output data portion and the training data records are used to train one of a plurality of predictive models that are each of a different predictive model type;
the training data records do not specify a predictive model type to be trained on the training data records; and
the predictive modeling server is configured to compare the output data portions of the training data records to a plurality of data formats, and based on the comparison, determining a training data type and a corresponding predictive model type to select a predictive model to train on the training data records;
receiving, at the client device and from the predictive modeling server system, access to a selected predictive model that has been trained on the training data records and that executes on the predictive modeling server;
sending, from the client device to the selected predicted model executing on the predictive modeling server, input data records and a prediction request; and
receiving, from the selected predictive model, predictive output generated by applying the received input data records to the selected predictive model.
14. The non-transitory computer-readable medium of claim 13 , wherein the training data records are sent through an HTTP web service.
15. The non-transitory computer-readable medium of claim 13 , wherein receiving access to the selected predictive model providing comprises receiving a URL address for accessing the selected predictive model.
16. The non-transitory computer-readable medium of claim 13 , wherein receiving access to the selected predictive model providing comprises receiving an interface configured to interface a trained model repository with a network communicatively coupled to the client device.
17. The non-transitory computer-readable medium of claim 16 , wherein the interface is further configured to interface a training data repository with the network communicatively coupled to the client device.
18. The non-transitory computer-readable medium of claim 13 , wherein the selected predictive model is selected at the predictive modeling server by:
generating a score for each of the plurality of trained predictive models, where each score represents an estimation of the effectiveness of the respective trained predictive model trained on the training data records; and
selecting a first trained predictive model from among the plurality of trained predictive models based on the generated scores.