Iris Classification

This example will take you through creating a simple prediction service to serve predictions for the Iris dataset. The predictive service will be deployed as a microservice inside Kubernetes allowing roll-back/up for scalable maintenance. It will be accessed via the Seldon server allowing you to run multiple algorithms and manage predictive services for many clients.


Create a Predictive Model

The first step is to create a predictive model based on some training data. In this case we have prepacked into 3 Docker images the process of creating a model for the Iris dataset using popular machine learning toolkits XGBoost, Vowpal Wabbit and Scikit-learn. Wrappers to call these libraries have been added to our python library to make integrating them as a microservice easy. However, you can build your model using any machine learning library.

The three example images are:

For details on building models using our python library see here.

Start a microservice

At runtime Seldon requires you expose your model scoring engine as a microservice API. In this example case the same image to create the models also exposes it for runtime scoring when run. We can start our chosen microservice using the command line script start-microservice

The script creates a Kubernetes deployment for the microservice in kubernetes/conf/microservices. If the microserice is already running Kubernetes will roll-down the previous version and roll-up the new version.

For example to start the XGBoost Iris microservice on the client “test” (created by seldon-up on startup):

start-microservice --type prediction --client test -i iris-xgboost seldonio/iris_xgboost:2.1 rest 1.0

Check with kubectl get pods -l name=iris-xgboost that the pod running the mircroservice is running.

Serve Predictions

You can now call the seldon server using the seldon CLI to test:

seldon-cli --quiet api --client-name test --endpoint /js/predict --json '{"data":{"f1":1,"f2":2.7,"f3":5.3,"f4":1.9}}'

The respone should be like:

  "meta": {
    "puid": "242fef28a255cd67d9c56ce6c15f6f0ff65d68e8",
    "modelName": "model_xgb",
    "variation": "iris-xgboost"
  "predictions": [
      "prediction": 0.00252304,
      "predictedClass": "Iris-setosa",
      "confidence": 0.00252304
      "prediction": 0.00350009,
      "predictedClass": "Iris-versicolor",
      "confidence": 0.00350009
      "prediction": 0.993977,
      "predictedClass": "Iris-virginica",
      "confidence": 0.993977
  "custom": null

Detailed Steps

The models for the Iris dataset are created using our pyseldon library to wrap calls to XGBoost and use Pandas along with a feature transform to normalise the features. The XGBoost version is shown below:

import sys, getopt, argparse
import seldon.pipeline.basic_transforms as bt
import seldon.pipeline.util as sutl
import seldon.pipeline.auto_transforms as pauto
from sklearn.pipeline import Pipeline
import seldon.xgb as xg
import sys

def run_pipeline(events,models):

    tNameId = bt.Feature_id_transform(min_size=0,exclude_missing=True,zero_based=True,input_feature="name",output_feature="nameId")
    tAuto = pauto.Auto_transform(max_values_numeric_categorical=2,exclude=["nameId","name"])
    xgb = xg.XGBoostClassifier(target="nameId",target_readable="name",excluded=["name"],learning_rate=0.1,silent=0)

    transformers = [("tName",tNameId),("tAuto",tAuto),("xgb",xgb)]
    p = Pipeline(transformers)

    pw = sutl.Pipeline_wrapper()
    df = pw.create_dataframe(events)
    df2 =

if __name__ == '__main__':
    parser = argparse.ArgumentParser(prog='xgb_pipeline')
    parser.add_argument('--events', help='events folder', required=True)
    parser.add_argument('--models', help='output models folder', required=True)

    args = parser.parse_args()
    opts = vars(args)


To run the runtime scorer for this model we load the model and using the pyseldon wrapper to start a simple Http server.

import sys, getopt, argparse
from seldon.microservice import Microservices

if __name__ == "__main__":
    import logging
    logger = logging.getLogger()
    logging.basicConfig(format='%(asctime)s : %(levelname)s : %(name)s : %(message)s', level=logging.DEBUG)

    parser = argparse.ArgumentParser(prog='microservice')
    parser.add_argument('--model_name', help='name of model', required=True)
    parser.add_argument('--pipeline', help='location of prediction pipeline', required=True)
    parser.add_argument('--aws_key', help='aws key', required=False)
    parser.add_argument('--aws_secret', help='aws secret', required=False)

    args = parser.parse_args()
    opts = vars(args)

    m = Microservices(aws_key=args.aws_key,aws_secret=args.aws_secret)
    app = m.create_prediction_microservice(args.pipeline,args.model_name)"", debug=False)

The full source code to create a docker image for model and runtime scorer for the three variants can be found in:

Next Steps