Here we can download the free dataset for Iris.

Iris dataset

Now we rename the file to csv and load it into Spark for check.

 scala> val schema = "sepal_length DOUBLE, sepal_width DOUBLE, petal_length DOUBLE,petal_width DOUBLE,species STRING"
 scala> val df = spark.read.schema(schema).csv("iris.csv")

This is a small dataset as you see below.

 scala> df.agg(countDistinct("species")).show
 +--------------+
 |count(species)|
 +--------------+
 |             3|
 +--------------+
 scala> df.groupBy("species").count.show
 +---------------+-----+
 |        species|count|
 +---------------+-----+
 | Iris-virginica|   50|
 |    Iris-setosa|   50|
 |Iris-versicolor|   50|
 +---------------+-----+

The column "species" is label and the other four columns are features. There are only three species for machine learning classification purpose.

So we write a script for this job, with sklearn as the ML library.

 import joblib
 import sklearn
 import pandas as pd
 from sklearn import svm
   
 #Read the training data from the file
 iris_data = pd.read_csv('./iris.csv',sep=',',names=["sepal_length", "sepal_width", "petal_length","petal_width","species"])
   
 #Assigning the classes and removing the target variable 
 iris_label = iris_data.pop('species')
   
 #We're going to be using the SVC (support vector classifier) SVM (support vector machine)
 classifier = svm.SVC(gamma='auto')
   
 #Training the model
 classifier.fit(iris_data, iris_label)
   
 #Saving the data locally
 model_filename = 'model.joblib'
 joblib.dump(classifier, model_filename)

As you see, only few lines of code we can train a mode to run this classification job.