mllib-classification-regression.md

layout: global
title: MLlib - Classification and Regression

• Table of contents {:toc}

\[ \newcommand{\R}{\mathbb{R}} \newcommand{\E}{\mathbb{E}} \newcommand{\x}{\mathbf{x}} \newcommand{\y}{\mathbf{y}} \newcommand{\wv}{\mathbf{w}} \newcommand{\av}{\mathbf{\alpha}} \newcommand{\bv}{\mathbf{b}} \newcommand{\N}{\mathbb{N}} \newcommand{\id}{\mathbf{I}} \newcommand{\ind}{\mathbf{1}} \newcommand{\0}{\mathbf{0}} \newcommand{\unit}{\mathbf{e}} \newcommand{\one}{\mathbf{1}} \newcommand{\zero}{\mathbf{0}} \]

Supervised Machine Learning

Supervised machine learning is the setting where we are given a set of training data examples $\{\x_i\}$, each example $\x_i$ coming with a corresponding label $y_i$. Given the training data $\{(\x_i,y_i)\}$, we want to learn a function to predict these labels. The two most well known classes of methods are classification, and regression. In classification, the label is a category (e.g. whether or not emails are spam), whereas in regression, the label is real value, and we want our prediction to be as close to the true value as possible.

Supervised Learning involves executing a learning Algorithm on a set of labeled training examples. The algorithm returns a trained Model (such as for example a linear function) that can predict the label for new data examples for which the label is unknown.

Discriminative Training using Linear Methods

Mathematical Formulation

Many standard machine learning methods can be formulated as a convex optimization problem, i.e. the task of finding a minimizer of a convex function $f$ that depends on a variable vector $\wv$ (called weights in the code), which has $d$ entries. Formally, we can write this as the optimization problem $\min_{\wv \in\R^d} \; f(\wv)$, where the objective function is of the form \begin{equation} f(\wv) := \lambda\, R(\wv) + \frac1n \sum_{i=1}^n L(\wv;\x_i,y_i) \label{eq:regPrimal} \ . \end{equation} Here the vectors $\x_i\in\R^d$ are the training data examples, for $1\le i\le n$, and $y_i\in\R$ are their corresponding labels, which we want to predict.

The objective function $f$ has two parts: The loss-function measures the error of the model on the training data. The loss-function $L(\wv;.)$ must be a convex function in $\wv$. The purpose of the regularizer is to encourage simple models, by punishing the complexity of the model $\wv$, in order to e.g. avoid over-fitting. Usually, the regularizer $R(.)$ is chosen as either the standard (Euclidean) L2-norm, $R(\wv) := \frac{1}{2}\|\wv\|^2$, or the L1-norm, $R(\wv) := \|\wv\|_1$, see below for more details.

The fixed regularization parameter $\lambda\ge0$ (regParam in the code) defines the trade-off between the two goals of small loss and small model complexity.

Binary Classification

Input: Datapoints $\x_i\in\R^{d}$, labels $y_i\in\{+1,-1\}$, for $1\le i\le n$.

Distributed Datasets. For all currently implemented optimization methods for classification, the data must be distributed between processes on the worker machines by examples. Machines hold consecutive blocks of the $n$ example/label pairs $(\x_i,y_i)$. In other words, the input distributed dataset (RDD) must be the set of vectors $\x_i\in\R^d$.

Support Vector Machine

The linear Support Vector Machine (SVM) has become a standard choice for classification tasks. Here the loss function in formulation $\eqref{eq:regPrimal}$ is given by the hinge-loss \[ L(\wv;\x_i,y_i) := \max \{0, 1-y_i \wv^T \x_i \} \ . \]

By default, SVMs are trained with an L2 regularization, which gives rise to the large-margin interpretation if these classifiers. We also support alternative L1 regularization. In this case, the primal optimization problem becomes an LP.

Logistic Regression

Despite its name, Logistic Regression is a binary classification method, again when the labels are given by binary values $y_i\in\{+1,-1\}$. The logistic loss function in formulation $\eqref{eq:regPrimal}$ is defined as \[ L(\wv;\x_i,y_i) := \log(1+\exp( -y_i \wv^T \x_i)) \ . \]

Linear Regression (Least Squares, Lasso and Ridge Regression)

Input: Data matrix $A\in\R^{n\times d}$, right hand side vector $\y\in\R^n$.

Distributed Datasets. For all currently implemented optimization methods for regression, the data matrix $A\in\R^{n\times d}$ must be distributed between the worker machines by rows of $A$. In other words, the input distributed dataset (RDD) must be the set of the $n$ rows $A_{i:}$ of $A$.

Least Squares Regression refers to the setting where we try to fit a vector $\y\in\R^n$ by linear combination of our observed data $A\in\R^{n\times d}$, which is given as a matrix.

It comes in 3 flavors:

Least Squares

Plain old least squares linear regression is the problem of minimizing \[ f_{\text{LS}}(\wv) := \frac1n \|A\wv-\y\|_2^2 \ . \]

Lasso

The popular Lasso (alternatively also known as $L_1$-regularized least squares regression) is given by \[ f_{\text{Lasso}}(\wv) := \frac1n \|A\wv-\y\|_2^2 + \lambda \|\wv\|_1 \ . \]

Ridge Regression

Ridge regression uses the same loss function but with a L2 regularizer term: \[ f_{\text{Ridge}}(\wv) := \frac1n \|A\wv-\y\|_2^2 + \frac{\lambda}{2}\|\wv\|^2 \ . \]

Loss Function. For all 3, the loss function (i.e. the measure of model fit) is given by the squared deviations from the right hand side $\y$. \[ \frac1n \|A\wv-\y\|_2^2 = \frac1n \sum_{i=1}^n (A_{i:} \wv - y_i )^2 = \frac1n \sum_{i=1}^n L(\wv;\x_i,y_i) \] This is also known as the mean squared error. In our generic problem formulation $\eqref{eq:regPrimal}$, this means the loss function is $L(\wv;\x_i,y_i) := (A_{i:} \wv - y_i )^2$, each depending only on a single row $A_{i:}$ of the data matrix $A$.

Using Different Regularizers

As we have mentioned above, the purpose of regularizer in $\eqref{eq:regPrimal}$ is to encourage simple models, by punishing the complexity of the model $\wv$, in order to e.g. avoid over-fitting. All machine learning methods for classification and regression that we have mentioned above are of interest for different types of regularization, the 3 most common ones being

• L2-Regularization. $R(\wv) := \frac{1}{2}\|\wv\|^2$. This regularizer is most commonly used for SVMs, logistic regression and ridge regression.

• L1-Regularization. $R(\wv) := \|\wv\|_1$. The L1 norm $\|\wv\|_1$ is the sum of the absolut values of the entries of a vector $\wv$. This regularizer is most commonly used for sparse methods, and feature selection, such as the Lasso.

• Non-Regularized. $R(\wv):=0$. Of course we can also train the models without any regularization, or equivalently by setting the regularization parameter $\lambda:=0$.

The optimization problems of the form $\eqref{eq:regPrimal}$ with convex regularizers such as the 3 mentioned here can be conveniently optimized with gradient descent type methods (such as SGD) which is implemented in MLlib currently, and explained in the next section.

Optimization Methods Working on the Primal Formulation

Stochastic subGradient Descent (SGD). For optimization objectives $f$ written as a sum, stochastic subgradient descent (SGD) can be an efficient choice of optimization method, as we describe in the optimization section in more detail. Because all methods considered here fit into the optimization formulation $\eqref{eq:regPrimal}$, this is especially natural, because the loss is written as an average of the individual losses coming from each datapoint.

Picking one datapoint $i\in[1..n]$ uniformly at random, we obtain a stochastic subgradient of $\eqref{eq:regPrimal}$, with respect to $\wv$ as follows: \[ f'_{\wv,i} := L'_{\wv,i} + \lambda\, R'_\wv \ , \] where $L'_{\wv,i} \in \R^d$ is a subgradient of the part of the loss function determined by the $i$-th datapoint, that is $L'_{\wv,i} \in \frac{\partial}{\partial \wv} L(\wv;\x_i,y_i)$. Furthermore, $R'_\wv$ is a subgradient of the regularizer $R(\wv)$, i.e. $R'_\wv \in \frac{\partial}{\partial \wv} R(\wv)$. The term $R'_\wv$ does not depend on which random datapoint is picked.

Gradients. The following table summarizes the gradients (or subgradients) of all loss functions and regularizers that we currently support:

	Function	Stochastic (Sub)Gradient
SVM Hinge Loss	$L(\wv;\x_i,y_i) := \max \{0, 1-y_i \wv^T \x_i \}$	$L'_{\wv,i} = \begin{cases}-y_i \x_i & \text{if $y_i \wv^T \x_i <1$}, \\ 0 & \text{otherwise}.\end{cases}$
Logistic Loss	$L(\wv;\x_i,y_i) := \log(1+\exp( -y_i \wv^T \x_i))$	$L'_{\wv,i} = -y_i \x_i \left(1-\frac1{1+\exp(-y_i \wv^T \x_i)} \right)$
Least Squares Loss	$L(\wv;\x_i,y_i) := (A_{i:} \wv - y_i)^2$	$L'_{\wv,i} = 2 A_{i:}^T (A_{i:} \wv - y_i)$
Non-Regularized	$R(\wv) := 0$	$R'_\wv = \0$
L2 Regularizer	$R(\wv) := \frac{1}{2}\|\wv\|^2$	$R'_\wv = \wv$
L1 Regularizer	$R(\wv) := \|\wv\|_1$	$R'_\wv = \mathop{sign}(\wv)$

Here $\mathop{sign}(\wv)$ is the vector consisting of the signs ($\pm1$) of all the entries of $\wv$. Also, note that $A_{i:} \in \R^d$ is a row-vector, but the gradient is a column vector.

Decision Tree Classification and Regression

Decision trees and their ensembles are popular methods for the machine learning tasks of classification and regression. Decision trees are widely used since they are easy to interpret, handle categorical variables, extend to the multi-class classification setting, do not require feature scaling and are able to capture non-linearities and feature interactions. Tree ensemble algorithms such as decision forest and boosting are among the top performers for classification and regression tasks.

Basic Algorithm

The decision tree is a greedy algorithm that performs a recursive binary partitioning of the feature space by choosing a single element from the best split set where each element of the set maximimizes the information gain at a tree node. In other words, the split chosen at each tree node is chosen from the set $\underset{s}{\operatorname{argmax}} IG(D,s)$ where $IG(D,s)$ is the information gain when a split $s$ is applied to a dataset $D$.

Node Impurity and Information Gain

The node impurity is a measure of the homogeneity of the labels at the node. The current implementation provides two impurity measures for classification (Gini index and entropy) and one impurity measure for regression (variance).

Impurity	Task	Formula	Description
Gini index	Classification	$\sum_{i=1}^{M} f_i(1-f_i)$	$f_i$ is the frequency of label $i$ at a node and $M$ is the number of unique labels.
Entropy	Classification	$\sum_{i=1}^{M} -f_ilog(f_i)$	$f_i$ is the frequency of label $i$ at a node and $M$ is the number of unique labels.
Variance	Classification	$\frac{1}{n} \sum_{i=1}^{N} (x_i - \mu)^2$	$y_i$ is label for an instance, $N$ is the number of instances and $\mu$ is the mean given by $\frac{1}{N} \sum_{i=1}^n x_i$.

The information gain is the difference in the parent node impurity and the weighted sum of the two child node impurities. Assuming that a split s partitions the dataset $D$ of size $N$ into two datasets $D_{left}$ and $D_{right}$ of sizes $N_{left}$ and $N_{right}$, respectively:

$IG(D,s) = Impurity(D) - \frac{N_{left}}{N} Impurity(D_{left}) - \frac{N_{right}}{N} Impurity(D_{right})$

Split Candidates

Continuous Features

For small datasets in single machine implementations, the split candidates for each continuous feature are typically the unique values for the feature. Some implementations sort the feature values and then use the ordered unique values as split candidates for faster tree calculations.

Finding ordered unique feature values is computationally intensive for large distributed datasets. One can get an approximate set of split candidates by performing a quantile calculation over a sampled fraction of the data. The ordered splits create "bins" and the maximum number of such bins can be specified using the maxBins parameters.

Note that the number of bins cannot be greater than the number of instances $N$ (a rare scenario since the default maxBins value is 100). The tree algorithm automatically reduces the number of bins if the condition is not satisfied.

Categorical Features

For $M$ categorical features, one could come up with $2^M-1$ split candidates. However, for binary classification, the number of split candidates can be reduced to $M-1$ by ordering the categorical feature values by the proportion of labels falling in one of the two classes (see Section 9.2.4 in Elements of Statistical Machine Learning for details). For example, for a binary classification problem with one categorical feature with three categories A, B and C with corresponding proportion of label 1 as 0.2, 0.6 and 0.4, the categorical features are orded as A followed by C followed B or A, B, C. The two split candidates are A | C, B and A , B | C where | denotes the split.

Stopping Rule

The recursive tree construction is stopped at a node when one of the two conditions is met:

1. The node depth is equal to the maxDepth training paramemter
2. No split candidate leads to an information gain at the node.

Practical Limitations

The tree implementation stores an Array[Double] of size O(#features * #splits * 2^maxDepth) in memory for aggregating histograms over partitions. The current implementation might not scale to very deep trees since the memory requirement grows exponentially with tree depth.

Please drop us a line if you encounter any issues. We are planning to solve this problem in the near future and real-world examples will be great.

Implementation in MLlib

Linear Methods

For both classification and regression algorithms with convex loss functions, MLlib implements a simple distributed version of stochastic subgradient descent (SGD), building on the underlying gradient descent primitive (as described in the optimization section). All provided algorithms take as input a regularization parameter (regParam) along with various parameters associated with stochastic gradient descent (stepSize, numIterations, miniBatchFraction). For each of them, we support all 3 possible regularizations (none, L1 or L2).

Available algorithms for binary classification:

• SVMWithSGD
• LogisticRegressionWithSGD

Available algorithms for linear regression:

• LinearRegressionWithSGD
• RidgeRegressionWithSGD
• LassoWithSGD

Behind the scenes, all above methods use the SGD implementation from the gradient descent primitive in MLlib, see the optimization part:

• GradientDescent

Tree-based Methods

The decision tree algorithm supports binary classification and regression:

• DecisionTee

Usage in Scala

Following code snippets can be executed in spark-shell.

Linear Methods

Binary Classification

The following code snippet illustrates how to load a sample dataset, execute a training algorithm on this training data using a static method in the algorithm object, and make predictions with the resulting model to compute the training error.

{% highlight scala %} import org.apache.spark.SparkContext import org.apache.spark.mllib.classification.SVMWithSGD import org.apache.spark.mllib.regression.LabeledPoint import org.apache.spark.mllib.linalg.Vectors

// Load and parse the data file val data = sc.textFile("mllib/data/sample_svm_data.txt") val parsedData = data.map { line => val parts = line.split(' ').map(_.toDouble) LabeledPoint(parts(0), Vectors.dense(parts.tail)) }

// Run training algorithm to build the model val numIterations = 100 val model = SVMWithSGD.train(parsedData, numIterations)

// Evaluate model on training examples and compute training error val labelAndPreds = parsedData.map { point => val prediction = model.predict(point.features) (point.label, prediction) } val trainErr = labelAndPreds.filter(r => r._1 != r._2).count.toDouble / parsedData.count println("Training Error = " + trainErr) {% endhighlight %}

The SVMWithSGD.train() method by default performs L2 regularization with the regularization parameter set to 1.0. If we want to configure this algorithm, we can customize SVMWithSGD further by creating a new object directly and calling setter methods. All other MLlib algorithms support customization in this way as well. For example, the following code produces an L1 regularized variant of SVMs with regularization parameter set to 0.1, and runs the training algorithm for 200 iterations.

{% highlight scala %} import org.apache.spark.mllib.optimization.L1Updater

val svmAlg = new SVMWithSGD() svmAlg.optimizer.setNumIterations(200) .setRegParam(0.1) .setUpdater(new L1Updater) val modelL1 = svmAlg.run(parsedData) {% endhighlight %}

Linear Regression

The following example demonstrate how to load training data, parse it as an RDD of LabeledPoint. The example then uses LinearRegressionWithSGD to build a simple linear model to predict label values. We compute the Mean Squared Error at the end to evaluate goodness of fit.

{% highlight scala %} import org.apache.spark.mllib.regression.LinearRegressionWithSGD import org.apache.spark.mllib.regression.LabeledPoint import org.apache.spark.mllib.linalg.Vectors

// Load and parse the data val data = sc.textFile("mllib/data/ridge-data/lpsa.data") val parsedData = data.map { line => val parts = line.split(',') LabeledPoint(parts(0).toDouble, Vectors.dense(parts(1).split(' ').map(_.toDouble))) }

// Building the model val numIterations = 100 val model = LinearRegressionWithSGD.train(parsedData, numIterations)

// Evaluate model on training examples and compute training error val valuesAndPreds = parsedData.map { point => val prediction = model.predict(point.features) (point.label, prediction) } val MSE = valuesAndPreds.map{case(v, p) => math.pow((v - p), 2)}.reduce(_ + _) / valuesAndPreds.count println("training Mean Squared Error = " + MSE) {% endhighlight %}

Similarly you can use RidgeRegressionWithSGD and LassoWithSGD and compare training Mean Squared Errors.

Decision Tree

Classification

The example below demonstrates how to load a CSV file, parse it as an RDD of LabeledPoint and then perform classification using a decision tree using Gini index as an impurity measure and a maximum tree depth of 5. The training error is calculated to measure the algorithm accuracy.

{% highlight scala %} import org.apache.spark.SparkContext import org.apache.spark.mllib.tree.DecisionTree import org.apache.spark.mllib.regression.LabeledPoint import org.apache.spark.mllib.linalg.Vectors import org.apache.spark.mllib.tree.configuration.Algo._ import org.apache.spark.mllib.tree.impurity.Gini

// Load and parse the data file val data = sc.textFile("mllib/data/sample_tree_data.csv") val parsedData = data.map { line => val parts = line.split(',').map(_.toDouble) LabeledPoint(parts(0), Vectors.dense(parts.tail)) }

// Run training algorithm to build the model val maxDepth = 5 val model = DecisionTree.train(parsedData, Classification, Gini, maxDepth)

// Evaluate model on training examples and compute training error val labelAndPreds = parsedData.map { point => val prediction = model.predict(point.features) (point.label, prediction) } val trainErr = labelAndPreds.filter(r => r._1 != r._2).count.toDouble / parsedData.count println("Training Error = " + trainErr) {% endhighlight %}

Regression

The example below demonstrates how to load a CSV file, parse it as an RDD of LabeledPoint and then perform regression using a decision tree using variance as an impurity measure and a maximum tree depth of 5. The Mean Squared Error is computed at the end to evaluate goodness of fit.

{% highlight scala %} import org.apache.spark.SparkContext import org.apache.spark.mllib.tree.DecisionTree import org.apache.spark.mllib.regression.LabeledPoint import org.apache.spark.mllib.linalg.Vectors import org.apache.spark.mllib.tree.configuration.Algo._ import org.apache.spark.mllib.tree.impurity.Variance

// Load and parse the data file val data = sc.textFile("mllib/data/sample_tree_data.csv") val parsedData = data.map { line => val parts = line.split(',').map(_.toDouble) LabeledPoint(parts(0), Vectors.dense(parts.tail)) }

// Run training algorithm to build the model val maxDepth = 5 val model = DecisionTree.train(parsedData, Regression, Variance, maxDepth)

// Evaluate model on training examples and compute training error val valuesAndPreds = parsedData.map { point => val prediction = model.predict(point.features) (point.label, prediction) } val MSE = valuesAndPreds.map{ case(v, p) => math.pow((v - p), 2)}.reduce(_ + _)/valuesAndPreds.count println("training Mean Squared Error = " + MSE) {% endhighlight %}

Usage in Java

All of MLlib's methods use Java-friendly types, so you can import and call them there the same way you do in Scala. The only caveat is that the methods take Scala RDD objects, while the Spark Java API uses a separate JavaRDD class. You can convert a Java RDD to a Scala one by calling .rdd() on your JavaRDD object.

Usage in Python

Following examples can be tested in the PySpark shell.

Linear Methods

Binary Classification

The following example shows how to load a sample dataset, build Logistic Regression model, and make predictions with the resulting model to compute the training error.

{% highlight python %} from pyspark.mllib.classification import LogisticRegressionWithSGD from pyspark.mllib.regression import LabeledPoint from numpy import array

Load and parse the data

def parsePoint(line): values = [float(x) for x in line.split(' ')] return LabeledPoint(values[0], values[1:])

data = sc.textFile("mllib/data/sample_svm_data.txt") parsedData = data.map(parsePoint)

Build the model

model = LogisticRegressionWithSGD.train(parsedData)

Evaluating the model on training data

labelsAndPreds = parsedData.map(lambda p: (p.label, model.predict(p.features))) trainErr = labelsAndPreds.filter(lambda (v, p): v != p).count() / float(parsedData.count()) print("Training Error = " + str(trainErr)) {% endhighlight %}

Linear Regression

The following example demonstrate how to load training data, parse it as an RDD of LabeledPoint. The example then uses LinearRegressionWithSGD to build a simple linear model to predict label values. We compute the Mean Squared Error at the end to evaluate goodness of fit.

{% highlight python %} from pyspark.mllib.regression import LabeledPoint, LinearRegressionWithSGD from numpy import array

Load and parse the data

def parsePoint(line): values = [float(x) for x in line.replace(',', ' ').split(' ')] return LabeledPoint(values[0], values[1:])

data = sc.textFile("mllib/data/ridge-data/lpsa.data") parsedData = data.map(parsePoint)

Build the model

model = LinearRegressionWithSGD.train(parsedData)

Evaluate the model on training data

valuesAndPreds = parsedData.map(lambda p: (p.label, model.predict(p.features))) MSE = valuesAndPreds.map(lambda (v, p): (v - p)**2).reduce(lambda x, y: x + y) / valuesAndPreds.count() print("Mean Squared Error = " + str(MSE)) {% endhighlight %}