Beginner's Guide for Python Users


Introduction

SystemML enables flexible, scalable machine learning. This flexibility is achieved through the specification of a high-level declarative machine learning language that comes in two flavors, one with an R-like syntax (DML) and one with a Python-like syntax (PyDML).

Algorithm scripts written in DML and PyDML can be run on Hadoop, on Spark, or in Standalone mode. No script modifications are required to change between modes. SystemML automatically performs advanced optimizations based on data and cluster characteristics, so much of the need to manually tweak algorithms is largely reduced or eliminated. To understand more about DML and PyDML, we recommend that you read Beginner’s Guide to DML and PyDML.

For convenience of Python users, SystemML exposes several language-level APIs that allow Python users to use SystemML and its algorithms without the need to know DML or PyDML. We explain these APIs in the below sections with example usecases.

Download & Setup

Before you get started on SystemML, make sure that your environment is set up and ready to go.

Install Java (need Java 8) and Apache Spark

If you already have an Apache Spark installation, you can skip this step.

/usr/bin/ruby -e "$(curl -fsSL https://raw.githubusercontent.com/Homebrew/install/master/install)"
brew tap caskroom/versions
brew cask install java8
brew install apache-spark

ruby -e "$(curl -fsSL https://raw.githubusercontent.com/Linuxbrew/install/master/install)"
brew tap caskroom/versions
brew cask install java8
brew install apache-spark

Install SystemML

To install released SystemML, please use following commands:

pip install systemml

pip3 install systemml

If you want to try out the bleeding edge version, please use following commands:

git checkout https://github.com/apache/systemml.git
cd systemml
mvn clean package -P distribution
pip install target/systemml-1.0.0-SNAPSHOT-python.tar.gz

git checkout https://github.com/apache/systemml.git
cd systemml
mvn clean package -P distribution
pip3 install target/systemml-1.0.0-SNAPSHOT-python.tar.gz

Uninstall SystemML

To uninstall SystemML, please use following command:

pip uninstall systemml

pip3 uninstall systemml

Start Pyspark shell

pyspark

PYSPARK_PYTHON=python3 pyspark

Matrix operations

To get started with SystemML, let’s try few elementary matrix multiplication operations:

import systemml as sml
import numpy as np
m1 = sml.matrix(np.ones((3,3)) + 2)
m2 = sml.matrix(np.ones((3,3)) + 3)
m2 = m1 * (m2 + m1)
m4 = 1.0 - m2
m4.sum(axis=1).toNumPy()

Output:

array([[-60.],
       [-60.],
       [-60.]])

Let us now write a simple script to train linear regression model: $ \beta = solve(X^T X, X^T y) $. For simplicity, we will use direct-solve method and ignore regularization parameter as well as intercept.

import numpy as np
from sklearn import datasets
import systemml as sml
# Load the diabetes dataset
diabetes = datasets.load_diabetes()
# Use only one feature
diabetes_X = diabetes.data[:, np.newaxis, 2]
# Split the data into training/testing sets
X_train = diabetes_X[:-20]
X_test = diabetes_X[-20:]
# Split the targets into training/testing sets
y_train = diabetes.target[:-20]
y_test = diabetes.target[-20:]
# Train Linear Regression model
X = sml.matrix(X_train)
y = sml.matrix(np.matrix(y_train).T)
A = X.transpose().dot(X)
b = X.transpose().dot(y)
beta = sml.solve(A, b).toNumPy()
y_predicted = X_test.dot(beta)
print('Residual sum of squares: %.2f' % np.mean((y_predicted - y_test) ** 2))

Output:

Residual sum of squares: 25282.12

We can improve the residual error by adding an intercept and regularization parameter. To do so, we will use mllearn API described in the next section.

Invoke SystemML’s algorithms

SystemML also exposes a subpackage mllearn. This subpackage allows Python users to invoke SystemML algorithms using Scikit-learn or MLPipeline API.

Scikit-learn interface

In the below example, we invoke SystemML’s Linear Regression algorithm.

import numpy as np
from sklearn import datasets
from systemml.mllearn import LinearRegression
# Load the diabetes dataset
diabetes = datasets.load_diabetes()
# Use only one feature
diabetes_X = diabetes.data[:, np.newaxis, 2]
# Split the data into training/testing sets
X_train = diabetes_X[:-20]
X_test = diabetes_X[-20:]
# Split the targets into training/testing sets
y_train = diabetes.target[:-20]
y_test = diabetes.target[-20:]
# Create linear regression object
regr = LinearRegression(spark, fit_intercept=True, C=float("inf"), solver='direct-solve')
# Train the model using the training sets
regr.fit(X_train, y_train)
y_predicted = regr.predict(X_test)
print('Residual sum of squares: %.2f' % np.mean((y_predicted - y_test) ** 2))

Output:

Residual sum of squares: 6991.17

As expected, by adding intercept and regularizer the residual error drops significantly.

Here is another example that where we invoke SystemML’s Logistic Regression algorithm on digits datasets.

# Scikit-learn way
from sklearn import datasets, neighbors
from systemml.mllearn import LogisticRegression
digits = datasets.load_digits()
X_digits = digits.data
y_digits = digits.target
n_samples = len(X_digits)
X_train = X_digits[:int(.9 * n_samples)]
y_train = y_digits[:int(.9 * n_samples)]
X_test = X_digits[int(.9 * n_samples):]
y_test = y_digits[int(.9 * n_samples):]
logistic = LogisticRegression(spark)
print('LogisticRegression score: %f' % logistic.fit(X_train, y_train).score(X_test, y_test))

Output:

LogisticRegression score: 0.927778

You can also save the trained model and load it later for prediction:

# Assuming logistic.fit(X_train, y_train) is already invoked
logistic.save('logistic_model')
new_logistic = LogisticRegression(spark)
new_logistic.load('logistic_model')
print('LogisticRegression score: %f' % new_logistic.score(X_test, y_test))

Passing PySpark DataFrame

To train the above algorithm on larger dataset, we can load the dataset into DataFrame and pass it to the fit method:

from sklearn import datasets
from systemml.mllearn import LogisticRegression
import pandas as pd
from sklearn.metrics import accuracy_score
import systemml as sml
digits = datasets.load_digits()
X_digits = digits.data
y_digits = digits.target
n_samples = len(X_digits)
# Split the data into training/testing sets and convert to PySpark DataFrame
df_train = sml.convertToLabeledDF(sqlCtx, X_digits[:int(.9 * n_samples)], y_digits[:int(.9 * n_samples)])
X_test = spark.createDataFrame(pd.DataFrame(X_digits[int(.9 * n_samples):]))
logistic = LogisticRegression(spark)
logistic.fit(df_train)
y_predicted = logistic.predict(X_test)
y_predicted = y_predicted.select('prediction').toPandas().as_matrix().flatten()
y_test = y_digits[int(.9 * n_samples):]
print('LogisticRegression score: %f' % accuracy_score(y_test, y_predicted))

Output:

LogisticRegression score: 0.922222

MLPipeline interface

In the below example, we demonstrate how the same LogisticRegression class can allow SystemML to fit seamlessly into large data pipelines.

# MLPipeline way
from pyspark.ml import Pipeline
from systemml.mllearn import LogisticRegression
from pyspark.ml.feature import HashingTF, Tokenizer
training = spark.createDataFrame([
    (0, "a b c d e spark", 1.0),
    (1, "b d", 2.0),
    (2, "spark f g h", 1.0),
    (3, "hadoop mapreduce", 2.0),
    (4, "b spark who", 1.0),
    (5, "g d a y", 2.0),
    (6, "spark fly", 1.0),
    (7, "was mapreduce", 2.0),
    (8, "e spark program", 1.0),
    (9, "a e c l", 2.0),
    (10, "spark compile", 1.0),
    (11, "hadoop software", 2.0)
], ["id", "text", "label"])
tokenizer = Tokenizer(inputCol="text", outputCol="words")
hashingTF = HashingTF(inputCol="words", outputCol="features", numFeatures=20)
lr = LogisticRegression(sqlCtx)
pipeline = Pipeline(stages=[tokenizer, hashingTF, lr])
model = pipeline.fit(training)
test = spark.createDataFrame([
    (12, "spark i j k"),
    (13, "l m n"),
    (14, "mapreduce spark"),
    (15, "apache hadoop")], ["id", "text"])
prediction = model.transform(test)
prediction.show()

Output:

+-------+---+---------------+------------------+--------------------+--------------------+----------+
|__INDEX| id|           text|             words|            features|         probability|prediction|
+-------+---+---------------+------------------+--------------------+--------------------+----------+
|    1.0| 12|    spark i j k|  [spark, i, j, k]|(20,[5,6,7],[2.0,...|[0.99999999999975...|       1.0|
|    2.0| 13|          l m n|         [l, m, n]|(20,[8,9,10],[1.0...|[1.37552128844736...|       2.0|
|    3.0| 14|mapreduce spark|[mapreduce, spark]|(20,[5,10],[1.0,1...|[0.99860290938153...|       1.0|
|    4.0| 15|  apache hadoop|  [apache, hadoop]|(20,[9,14],[1.0,1...|[5.41688748236143...|       2.0|
+-------+---+---------------+------------------+--------------------+--------------------+----------+

Invoking DML/PyDML scripts using MLContext

The below example demonstrates how to invoke the algorithm scripts/algorithms/MultiLogReg.dml using Python MLContext API.

from sklearn import datasets
from pyspark.sql import SQLContext
import systemml as sml
import pandas as pd
digits = datasets.load_digits()
X_digits = digits.data
y_digits = digits.target + 1
n_samples = len(X_digits)
# Split the data into training/testing sets and convert to PySpark DataFrame
X_df = sqlCtx.createDataFrame(pd.DataFrame(X_digits[:int(.9 * n_samples)]))
y_df = sqlCtx.createDataFrame(pd.DataFrame(y_digits[:int(.9 * n_samples)]))
ml = sml.MLContext(sc)
# Run the MultiLogReg.dml script at the given URL
scriptUrl = "https://raw.githubusercontent.com/apache/systemml/master/scripts/algorithms/MultiLogReg.dml"
script = sml.dml(scriptUrl).input(X=X_df, Y_vec=y_df).output("B_out")
beta = ml.execute(script).get('B_out').toNumPy()