Vowpal Wabbit on Azure ML

Example to run Vowpal Wabbit on Azure ML

pre-requisites

• Azure account
• Azure machine learning Service

Steps

• Create a notebook with python 3.8 with Azure ML as kernel
• Install vowpal wabbit

pip install vowpalwabbit

• Load the necessary libraries

import pandas as pd
import sklearn
import sklearn.model_selection
import sklearn.datasets
import vowpalwabbit

• now load the iris data set into dataframe

iris_dataset = sklearn.datasets.load_iris()
iris_dataframe = pd.DataFrame(
    data=iris_dataset.data, columns=iris_dataset.feature_names
)
# vw expects labels starting from 1
iris_dataframe["y"] = iris_dataset.target + 1
training_data, testing_data = sklearn.model_selection.train_test_split(
    iris_dataframe, test_size=0.2
)

• Format feature function

def to_vw_format(row):
    res = f"{int(row.y)} |"
    for idx, value in row.drop(["y"]).iteritems():
        feature_name = idx.replace(" ", "_").replace("(", "").replace(")", "")
        res += f" {feature_name}:{value}"
    return res

• diplay few rows

for ex in training_data.head(10).apply(to_vw_format, axis=1):
    print(ex)

• Create vowpall workspace
• Run training with multiple samples

vw = vowpalwabbit.Workspace("--oaa 3 --quiet")
# learn from training set with multiple passes
for example in training_data.apply(to_vw_format, axis=1):
    vw.learn(example)
# predict from the testing set
predictions = []
for example in testing_data.apply(to_vw_format, axis=1):
    predicted_class = vw.predict(example)
    predictions.append(predicted_class)

• now Predict the output

accuracy = len(testing_data[testing_data.y == predictions]) / len(testing_data)
print(f"Model accuracy {accuracy}")

• output should be

Model accuracy 0.7333333333333333

Originally published at https://github.com.

Mlearning.ai Submission Suggestions