Model Training

Train a new model on a TFrame using Terrene's API

post
Get Algorithm Suggestion

https://<your-domain>.api.terrene.co/models/<object_id>/suggest_trainers/
Terrene will recommend a number of algorithms to train a model on based on the supplied training data.
Request
Response
Request
Path Parameters
object_id
required
string
ID of the predictive model
Headers
Authorization
required
string
Please see Authentication section for more information
Body Parameters
count
optional
number
Number of algorithms for Terrene to suggest
store
required
string
The ID of the training data TFrame
Response
200: OK
Suggestions are ranked by viability in the "suggestions" attribute.
"suggestions":[
"RandomForestRegressor",
"KNeighborsRegressor",
"NeuralNetworkRegressor",
"ProbabilisticModel",
"BayesianRidge"
],
"settings":{
"RandomForestRegressor":[
],
"KNeighborsRegressor":[
],
"NeuralNetworkRegressor":[
{
"name":"optimizer",
"default":"adam",
"options":[
"adam",
"rmsprop",
"adagrad",
"sgd",
"nadam"
]
},
{
"name":"activation",
"default":"reul",
"options":[
"relu",
"tanh",
"softmax",
"elu",
"softplus",
"sigmoid"
]
},
{
"name":"sub_models_count",
"default":5,
"range":[
1,
10
]
},
{
"name":"epoch_per_sub_model",
"default":30,
"range":[
10,
100
]
},
{
"name":"validation_split",
"default":0.1,
"options":[
0.1,
0.2,
0.3
]
}
],
"ProbabilisticModel":[
{
"name":"optimizer",
"default":"adam",
"options":[
"adam",
"rmsprop",
"adagrad",
"sgd",
"nadam"
]
},
{
"name":"activation",
"default":"reul",
"options":[
"relu",
"tanh",
"softmax",
"elu",
"softplus",
"sigmoid"
]
},
{
"name":"sub_models_count",
"default":5,
"range":[
1,
10
]
},
{
"name":"epoch_per_sub_model",
"default":30,
"range":[
10,
100
]
},
{
"name":"validation_split",
"default":0.1,
"options":[
0.1,
0.2,
0.3
]
}
],
"BayesianRidge":[
]
}
}

Your API endpoint address is unique to your Terrene instance. To find out the API endpoint for your instance, please contact support@terrene.co

post
Train a Model

https://<your-domain>.api.terrene.co/models/<object_id>/train/
This endpoint allows you train a model with Terrene using your algorithm and training data of choice
Request
Response
Request
Path Parameters
object_id
required
string
ID of the predictive model
Headers
Authorization
required
string
Please see Authentication section for more information
Body Parameters
trainer
optional
number
Name of your choice of algorithm suggested by Terrene
store
required
string
The ID of the training data TFrame
Response
200: OK
Returns a background task with the status of the training job.
{
"object_id":"4051c542-27ba-43fd-b542-42c401ce1f04",
"created_at":"2019-12-16T06:36:54.027333Z",
"updated_at":"2019-12-16T06:36:54.080478Z",
"name":"Training dda474fe-ffd5-4fb3-bba4-92cdca2cbf14 2019-12-16 06:36:54.026929",
"progress":0,
"results":null,
"resources":[
{
"object_id":"2339677f-1648-40c8-a093-620ee5ec03d0",
"created_at":"2019-12-11T15:57:35.446205Z",
"updated_at":"2019-12-11T15:57:35.446269Z",
"name":"2019-07-11 Ontario Training Data",
"description":"Training data",
"tags":null,
"public":false,
"read_access":true,
"write_access":true,
"execute_access":true,
"metadata":{
}
},
{
"object_id":"dda474fe-ffd5-4fb3-bba4-92cdca2cbf14",
"created_at":"2019-12-11T15:06:49.657108Z",
"updated_at":"2019-12-11T15:06:49.657201Z",
"name":"sagacious moldmaker Predictive Model",
"description":"this model predicts ieso_forecast_error",
"tags":null,
"public":false,
"read_access":true,
"write_access":true,
"execute_access":true,
"metadata":{
}
}
],
"failed":false,
"running_time":0
}