Next step after training ml model?

Solved!
dhruvrawat101
Level 1
Next step after training ml model?

While using DSS, I took a dateset, prepared it, made some charts for visualization stuff, trained it using a ML model ,got certain predictions out of it with R2 score 0.95.

The question is what next?

I am using free version so i don't think deployment can be done in free version .Please help me to deploy it as a web based application so that i can showcase it to someone.

Please tell the next steps after training model ..

 

0 Kudos
1 Solution
Andrey
Dataiker Alumni

Hi,

Once the model is trained in the Lab, you can deploy it to the flow. From there you can either use it for batch scoring (by using a Score recipe and a test dataset) or by creating an API.

Please refer to this page for more info about API creation:

https://doc.dataiku.com/dss/latest/apinode/first-service-apideployer.html#create-the-api-directly-fr...

For example, once the endpoint is created in the design node, you can send the following request:

curl --location --request POST 'http://localhost:4750/public/api/v1/service_id/entrypoint_id/predict' \
--header 'Content-Type: application/json' \
--data-raw '{
"features": {
"PassengerId": "892",
"Pclass": "3",
"Name": "Kelly, Mr. James",
"Sex": "male",
"Age": "34.5",
"SibSp": "0",
"Parch": "0",
"Ticket": "330911",
"Fare": "7.8292",
"Embarked": "Q"
}
}'

Where the port can be found by clicking on "Actions":

Screenshot 2020-07-06 at 12.04.11.png

 

service_id and entrypoint_id is whatever you specify at the creation and data is the JSON you want to score (in my case it's the Titanic dataset row)

 

Andrey Avtomonov
R&D Engineer @ Dataiku

View solution in original post

0 Kudos
1 Reply
Andrey
Dataiker Alumni

Hi,

Once the model is trained in the Lab, you can deploy it to the flow. From there you can either use it for batch scoring (by using a Score recipe and a test dataset) or by creating an API.

Please refer to this page for more info about API creation:

https://doc.dataiku.com/dss/latest/apinode/first-service-apideployer.html#create-the-api-directly-fr...

For example, once the endpoint is created in the design node, you can send the following request:

curl --location --request POST 'http://localhost:4750/public/api/v1/service_id/entrypoint_id/predict' \
--header 'Content-Type: application/json' \
--data-raw '{
"features": {
"PassengerId": "892",
"Pclass": "3",
"Name": "Kelly, Mr. James",
"Sex": "male",
"Age": "34.5",
"SibSp": "0",
"Parch": "0",
"Ticket": "330911",
"Fare": "7.8292",
"Embarked": "Q"
}
}'

Where the port can be found by clicking on "Actions":

Screenshot 2020-07-06 at 12.04.11.png

 

service_id and entrypoint_id is whatever you specify at the creation and data is the JSON you want to score (in my case it's the Titanic dataset row)

 

Andrey Avtomonov
R&D Engineer @ Dataiku
0 Kudos

Labels

?
Labels (3)
A banner prompting to get Dataiku