二十種特徵變換方法及Spark MLlib調用實例（Scala/Java/python）（一）

Tokenizer（分詞器）

算法介紹：

        Tokenization將文本劃分爲獨立個體（通常爲單詞）。下面的例子展示瞭如何把句子劃分爲單詞。

        RegexTokenizer基於正則表達式提供更多的劃分選項。默認情況下，參數“pattern”爲劃分文本的分隔符。或者，用戶可以指定參數“gaps”來指明正則“patten”表示“tokens”而不是分隔符，這樣來爲分詞結果找到所有可能匹配的情況。

示例調用：

Scala：

import org.apache.spark.ml.feature.{RegexTokenizer, Tokenizer}

val sentenceDataFrame = spark.createDataFrame(Seq(
  (0, "Hi I heard about Spark"),
  (1, "I wish Java could use case classes"),
  (2, "Logistic,regression,models,are,neat")
)).toDF("label", "sentence")

val tokenizer = new Tokenizer().setInputCol("sentence").setOutputCol("words")
val regexTokenizer = new RegexTokenizer()
  .setInputCol("sentence")
  .setOutputCol("words")
  .setPattern("\\W") // alternatively .setPattern("\\w+").setGaps(false)

val tokenized = tokenizer.transform(sentenceDataFrame)
tokenized.select("words", "label").take(3).foreach(println)
val regexTokenized = regexTokenizer.transform(sentenceDataFrame)
regexTokenized.select("words", "label").take(3).foreach(println)

Java：

import java.util.Arrays;
import java.util.List;

import org.apache.spark.ml.feature.RegexTokenizer;
import org.apache.spark.ml.feature.Tokenizer;
import org.apache.spark.sql.Dataset;
import org.apache.spark.sql.Row;
import org.apache.spark.sql.RowFactory;
import org.apache.spark.sql.types.DataTypes;
import org.apache.spark.sql.types.Metadata;
import org.apache.spark.sql.types.StructField;
import org.apache.spark.sql.types.StructType;

List<Row> data = Arrays.asList(
  RowFactory.create(0, "Hi I heard about Spark"),
  RowFactory.create(1, "I wish Java could use case classes"),
  RowFactory.create(2, "Logistic,regression,models,are,neat")
);

StructType schema = new StructType(new StructField[]{
  new StructField("label", DataTypes.IntegerType, false, Metadata.empty()),
  new StructField("sentence", DataTypes.StringType, false, Metadata.empty())
});

Dataset<Row> sentenceDataFrame = spark.createDataFrame(data, schema);

Tokenizer tokenizer = new Tokenizer().setInputCol("sentence").setOutputCol("words");

Dataset<Row> wordsDataFrame = tokenizer.transform(sentenceDataFrame);
for (Row r : wordsDataFrame.select("words", "label").takeAsList(3)) {
  java.util.List<String> words = r.getList(0);
  for (String word : words) System.out.print(word + " ");
  System.out.println();
}

RegexTokenizer regexTokenizer = new RegexTokenizer()
  .setInputCol("sentence")
  .setOutputCol("words")
  .setPattern("\\W");  // alternatively .setPattern("\\w+").setGaps(false);

Python：

from pyspark.ml.feature import Tokenizer, RegexTokenizer

sentenceDataFrame = spark.createDataFrame([
    (0, "Hi I heard about Spark"),
    (1, "I wish Java could use case classes"),
    (2, "Logistic,regression,models,are,neat")
], ["label", "sentence"])
tokenizer = Tokenizer(inputCol="sentence", outputCol="words")
wordsDataFrame = tokenizer.transform(sentenceDataFrame)
for words_label in wordsDataFrame.select("words", "label").take(3):
    print(words_label)
regexTokenizer = RegexTokenizer(inputCol="sentence", outputCol="words", pattern="\\W")
# alternatively, pattern="\\w+", gaps(False)

StopWordsRemover

算法介紹：

       停用詞爲在文檔中頻繁出現，但未承載太多意義的詞語，他們不應該被包含在算法輸入中。

       StopWordsRemover的輸入爲一系列字符串（如分詞器輸出），輸出中刪除了所有停用詞。停用詞表由stopWords參數提供。一些語言的默認停用詞表可以通過StopWordsRemover.loadDefaultStopWords(language)調用。布爾參數caseSensitive指明是否區分大小寫（默認爲否）。

示例：

假設我們有如下DataFrame，有id和raw兩列：

id | raw

----|----------

 0  | [I,saw, the, red, baloon]

 1  |[Mary, had, a, little, lamb]

通過對raw列調用StopWordsRemover，我們可以得到篩選出的結果列如下：

id | raw                         | filtered

----|-----------------------------|--------------------

 0  | [I,saw, the, red, baloon]  |  [saw, red, baloon]

 1  |[Mary, had, a, little, lamb]|[Mary, little, lamb]

其中，“I”, “the”, “had”以及“a”被移除。

示例調用：

Scala：

import org.apache.spark.ml.feature.StopWordsRemover

val remover = new StopWordsRemover()
  .setInputCol("raw")
  .setOutputCol("filtered")

val dataSet = spark.createDataFrame(Seq(
  (0, Seq("I", "saw", "the", "red", "baloon")),
  (1, Seq("Mary", "had", "a", "little", "lamb"))
)).toDF("id", "raw")

remover.transform(dataSet).show()

Java：

import java.util.Arrays;
import java.util.List;

import org.apache.spark.ml.feature.StopWordsRemover;
import org.apache.spark.sql.Dataset;
import org.apache.spark.sql.Row;
import org.apache.spark.sql.RowFactory;
import org.apache.spark.sql.types.DataTypes;
import org.apache.spark.sql.types.Metadata;
import org.apache.spark.sql.types.StructField;
import org.apache.spark.sql.types.StructType;

StopWordsRemover remover = new StopWordsRemover()
  .setInputCol("raw")
  .setOutputCol("filtered");

List<Row> data = Arrays.asList(
  RowFactory.create(Arrays.asList("I", "saw", "the", "red", "baloon")),
  RowFactory.create(Arrays.asList("Mary", "had", "a", "little", "lamb"))
);

StructType schema = new StructType(new StructField[]{
  new StructField(
    "raw", DataTypes.createArrayType(DataTypes.StringType), false, Metadata.empty())
});

Dataset<Row> dataset = spark.createDataFrame(data, schema);
remover.transform(dataset).show();

Python：

from pyspark.ml.feature import StopWordsRemover

sentenceData = spark.createDataFrame([
    (0, ["I", "saw", "the", "red", "baloon"]),
    (1, ["Mary", "had", "a", "little", "lamb"])
], ["label", "raw"])

remover = StopWordsRemover(inputCol="raw", outputCol="filtered")
remover.transform(sentenceData).show(truncate=False)

n-gram

算法介紹：

       一個n-gram是一個長度爲整數n的字序列。NGram可以用來將輸入轉換爲n-gram。

        NGram的輸入爲一系列字符串（如分詞器輸出）。參數n決定每個n-gram包含的對象個數。結果包含一系列n-gram，其中每個n-gram代表一個空格分割的n個連續字符。如果輸入少於n個字符串，將沒有輸出結果。

示例調用：

Scala：

import org.apache.spark.ml.feature.NGram

val wordDataFrame = spark.createDataFrame(Seq(
  (0, Array("Hi", "I", "heard", "about", "Spark")),
  (1, Array("I", "wish", "Java", "could", "use", "case", "classes")),
  (2, Array("Logistic", "regression", "models", "are", "neat"))
)).toDF("label", "words")

val ngram = new NGram().setInputCol("words").setOutputCol("ngrams")
val ngramDataFrame = ngram.transform(wordDataFrame)
ngramDataFrame.take(3).map(_.getAs[Stream[String]]("ngrams").toList).foreach(println)

Java：

import java.util.Arrays;
import java.util.List;

import org.apache.spark.ml.feature.NGram;
import org.apache.spark.sql.Row;
import org.apache.spark.sql.RowFactory;
import org.apache.spark.sql.types.DataTypes;
import org.apache.spark.sql.types.Metadata;
import org.apache.spark.sql.types.StructField;
import org.apache.spark.sql.types.StructType;

List<Row> data = Arrays.asList(
  RowFactory.create(0.0, Arrays.asList("Hi", "I", "heard", "about", "Spark")),
  RowFactory.create(1.0, Arrays.asList("I", "wish", "Java", "could", "use", "case", "classes")),
  RowFactory.create(2.0, Arrays.asList("Logistic", "regression", "models", "are", "neat"))
);

StructType schema = new StructType(new StructField[]{
  new StructField("label", DataTypes.DoubleType, false, Metadata.empty()),
  new StructField(
    "words", DataTypes.createArrayType(DataTypes.StringType), false, Metadata.empty())
});

Dataset<Row> wordDataFrame = spark.createDataFrame(data, schema);

NGram ngramTransformer = new NGram().setInputCol("words").setOutputCol("ngrams");

Dataset<Row> ngramDataFrame = ngramTransformer.transform(wordDataFrame);

for (Row r : ngramDataFrame.select("ngrams", "label").takeAsList(3)) {
  java.util.List<String> ngrams = r.getList(0);
  for (String ngram : ngrams) System.out.print(ngram + " --- ");
  System.out.println();
}

Python：

from pyspark.ml.feature import NGram

wordDataFrame = spark.createDataFrame([
    (0, ["Hi", "I", "heard", "about", "Spark"]),
    (1, ["I", "wish", "Java", "could", "use", "case", "classes"]),
    (2, ["Logistic", "regression", "models", "are", "neat"])
], ["label", "words"])
ngram = NGram(inputCol="words", outputCol="ngrams")
ngramDataFrame = ngram.transform(wordDataFrame)
for ngrams_label in ngramDataFrame.select("ngrams", "label").take(3):
    print(ngrams_label)

Binarizer

算法介紹：

       二值化是根據閥值將連續數值特徵轉換爲0-1特徵的過程。

       Binarizer參數有輸入、輸出以及閥值。特徵值大於閥值將映射爲1.0，特徵值小於等於閥值將映射爲0.0。

示例調用：

Scala：

import org.apache.spark.ml.feature.Binarizer

val data = Array((0, 0.1), (1, 0.8), (2, 0.2))
val dataFrame = spark.createDataFrame(data).toDF("label", "feature")

val binarizer: Binarizer = new Binarizer()
  .setInputCol("feature")
  .setOutputCol("binarized_feature")
  .setThreshold(0.5)

val binarizedDataFrame = binarizer.transform(dataFrame)
val binarizedFeatures = binarizedDataFrame.select("binarized_feature")
binarizedFeatures.collect().foreach(println)

Java：

import java.util.Arrays;
import java.util.List;

import org.apache.spark.ml.feature.Binarizer;
import org.apache.spark.sql.Row;
import org.apache.spark.sql.RowFactory;
import org.apache.spark.sql.types.DataTypes;
import org.apache.spark.sql.types.Metadata;
import org.apache.spark.sql.types.StructField;
import org.apache.spark.sql.types.StructType;

List<Row> data = Arrays.asList(
  RowFactory.create(0, 0.1),
  RowFactory.create(1, 0.8),
  RowFactory.create(2, 0.2)
);
StructType schema = new StructType(new StructField[]{
  new StructField("id", DataTypes.IntegerType, false, Metadata.empty()),
  new StructField("feature", DataTypes.DoubleType, false, Metadata.empty())
});
Dataset<Row> continuousDataFrame = spark.createDataFrame(data, schema);
Binarizer binarizer = new Binarizer()
  .setInputCol("feature")
  .setOutputCol("binarized_feature")
  .setThreshold(0.5);
Dataset<Row> binarizedDataFrame = binarizer.transform(continuousDataFrame);
Dataset<Row> binarizedFeatures = binarizedDataFrame.select("binarized_feature");
for (Row r : binarizedFeatures.collectAsList()) {
  Double binarized_value = r.getDouble(0);
  System.out.println(binarized_value);
}

Python：

from pyspark.ml.feature import Binarizer

continuousDataFrame = spark.createDataFrame([
    (0, 0.1),
    (1, 0.8),
    (2, 0.2)
], ["label", "feature"])
binarizer = Binarizer(threshold=0.5, inputCol="feature", outputCol="binarized_feature")
binarizedDataFrame = binarizer.transform(continuousDataFrame)
binarizedFeatures = binarizedDataFrame.select("binarized_feature")
for binarized_feature, in binarizedFeatures.collect():
    print(binarized_feature)

PCA

算法介紹：

        主成分分析是一種統計學方法，它使用正交轉換從一系列可能相關的變量中提取線性無關變量集，提取出的變量集中的元素稱爲主成分。使用PCA方法可以對變量集合進行降維。下面的示例將會展示如何將5維特徵向量轉換爲3維主成分向量。

示例調用：

Scala：

import org.apache.spark.ml.feature.PCA
import org.apache.spark.ml.linalg.Vectors

val data = Array(
  Vectors.sparse(5, Seq((1, 1.0), (3, 7.0))),
  Vectors.dense(2.0, 0.0, 3.0, 4.0, 5.0),
  Vectors.dense(4.0, 0.0, 0.0, 6.0, 7.0)
)
val df = spark.createDataFrame(data.map(Tuple1.apply)).toDF("features")
val pca = new PCA()
  .setInputCol("features")
  .setOutputCol("pcaFeatures")
  .setK(3)
  .fit(df)
val pcaDF = pca.transform(df)
val result = pcaDF.select("pcaFeatures")
result.show()

Java：

import java.util.Arrays;
import java.util.List;

import org.apache.spark.ml.feature.PCA;
import org.apache.spark.ml.feature.PCAModel;
import org.apache.spark.ml.linalg.VectorUDT;
import org.apache.spark.ml.linalg.Vectors;
import org.apache.spark.sql.Dataset;
import org.apache.spark.sql.Row;
import org.apache.spark.sql.RowFactory;
import org.apache.spark.sql.types.Metadata;
import org.apache.spark.sql.types.StructField;
import org.apache.spark.sql.types.StructType;

List<Row> data = Arrays.asList(
  RowFactory.create(Vectors.sparse(5, new int[]{1, 3}, new double[]{1.0, 7.0})),
  RowFactory.create(Vectors.dense(2.0, 0.0, 3.0, 4.0, 5.0)),
  RowFactory.create(Vectors.dense(4.0, 0.0, 0.0, 6.0, 7.0))
);

StructType schema = new StructType(new StructField[]{
  new StructField("features", new VectorUDT(), false, Metadata.empty()),
});

Dataset<Row> df = spark.createDataFrame(data, schema);

PCAModel pca = new PCA()
  .setInputCol("features")
  .setOutputCol("pcaFeatures")
  .setK(3)
  .fit(df);

Dataset<Row> result = pca.transform(df).select("pcaFeatures");
result.show();

Python：

from pyspark.ml.feature import PCA
from pyspark.ml.linalg import Vectors

data = [(Vectors.sparse(5, [(1, 1.0), (3, 7.0)]),),
        (Vectors.dense([2.0, 0.0, 3.0, 4.0, 5.0]),),
        (Vectors.dense([4.0, 0.0, 0.0, 6.0, 7.0]),)]
df = spark.createDataFrame(data, ["features"])
pca = PCA(k=3, inputCol="features", outputCol="pcaFeatures")
model = pca.fit(df)
result = model.transform(df).select("pcaFeatures")
result.show(truncate=False)

PolynomialExpansion

算法介紹：

       多項式擴展通過產生n維組合將原始特徵將特徵擴展到多項式空間。下面的示例會介紹如何將你的特徵集拓展到3維多項式空間。

示例調用：

Scala：

import org.apache.spark.ml.feature.PolynomialExpansion
import org.apache.spark.ml.linalg.Vectors

val data = Array(
  Vectors.dense(-2.0, 2.3),
  Vectors.dense(0.0, 0.0),
  Vectors.dense(0.6, -1.1)
)
val df = spark.createDataFrame(data.map(Tuple1.apply)).toDF("features")
val polynomialExpansion = new PolynomialExpansion()
  .setInputCol("features")
  .setOutputCol("polyFeatures")
  .setDegree(3)
val polyDF = polynomialExpansion.transform(df)
polyDF.select("polyFeatures").take(3).foreach(println)

Java：

import java.util.Arrays;
import java.util.List;

import org.apache.spark.ml.feature.PolynomialExpansion;
import org.apache.spark.ml.linalg.VectorUDT;
import org.apache.spark.ml.linalg.Vectors;
import org.apache.spark.sql.Dataset;
import org.apache.spark.sql.Row;
import org.apache.spark.sql.RowFactory;
import org.apache.spark.sql.types.Metadata;
import org.apache.spark.sql.types.StructField;
import org.apache.spark.sql.types.StructType;

PolynomialExpansion polyExpansion = new PolynomialExpansion()
  .setInputCol("features")
  .setOutputCol("polyFeatures")
  .setDegree(3);

List<Row> data = Arrays.asList(
  RowFactory.create(Vectors.dense(-2.0, 2.3)),
  RowFactory.create(Vectors.dense(0.0, 0.0)),
  RowFactory.create(Vectors.dense(0.6, -1.1))
);

StructType schema = new StructType(new StructField[]{
  new StructField("features", new VectorUDT(), false, Metadata.empty()),
});

Dataset<Row> df = spark.createDataFrame(data, schema);
Dataset<Row> polyDF = polyExpansion.transform(df);

List<Row> rows = polyDF.select("polyFeatures").takeAsList(3);
for (Row r : rows) {
  System.out.println(r.get(0));
}

Python：

from pyspark.ml.feature import PolynomialExpansion
from pyspark.ml.linalg import Vectors

df = spark\
    .createDataFrame([(Vectors.dense([-2.0, 2.3]),),
                      (Vectors.dense([0.0, 0.0]),),
                      (Vectors.dense([0.6, -1.1]),)],
                     ["features"])
px = PolynomialExpansion(degree=3, inputCol="features", outputCol="polyFeatures")
polyDF = px.transform(df)
for expanded in polyDF.select("polyFeatures").take(3):
    print(expanded)

Discrete Cosine Transform(DCT)

算法介紹：

       離散餘弦變換是與傅里葉變換相關的一種變換，它類似於離散傅立葉變換但是隻使用實數。離散餘弦變換相當於一個長度大概是它兩倍的離散傅里葉變換，這個離散傅里葉變換是對一個實偶函數進行的（因爲一個實偶函數的傅里葉變換仍然是一個實偶函數）。離散餘弦變換，經常被信號處理和圖像處理使用，用於對信號和圖像（包括靜止圖像和運動圖像）進行有損數據壓縮。

示例調用：

Scala：

import org.apache.spark.ml.feature.DCT
import org.apache.spark.ml.linalg.Vectors

val data = Seq(
  Vectors.dense(0.0, 1.0, -2.0, 3.0),
  Vectors.dense(-1.0, 2.0, 4.0, -7.0),
  Vectors.dense(14.0, -2.0, -5.0, 1.0))

val df = spark.createDataFrame(data.map(Tuple1.apply)).toDF("features")

val dct = new DCT()
  .setInputCol("features")
  .setOutputCol("featuresDCT")
  .setInverse(false)

val dctDf = dct.transform(df)
dctDf.select("featuresDCT").show(3)

Java：

import java.util.Arrays;
import java.util.List;

import org.apache.spark.ml.feature.DCT;
import org.apache.spark.ml.linalg.VectorUDT;
import org.apache.spark.ml.linalg.Vectors;
import org.apache.spark.sql.Row;
import org.apache.spark.sql.RowFactory;
import org.apache.spark.sql.types.Metadata;
import org.apache.spark.sql.types.StructField;
import org.apache.spark.sql.types.StructType;

List<Row> data = Arrays.asList(
  RowFactory.create(Vectors.dense(0.0, 1.0, -2.0, 3.0)),
  RowFactory.create(Vectors.dense(-1.0, 2.0, 4.0, -7.0)),
  RowFactory.create(Vectors.dense(14.0, -2.0, -5.0, 1.0))
);
StructType schema = new StructType(new StructField[]{
  new StructField("features", new VectorUDT(), false, Metadata.empty()),
});
Dataset<Row> df = spark.createDataFrame(data, schema);
DCT dct = new DCT()
  .setInputCol("features")
  .setOutputCol("featuresDCT")
  .setInverse(false);
Dataset<Row> dctDf = dct.transform(df);
dctDf.select("featuresDCT").show(3);

Python：

from pyspark.ml.feature import DCT
from pyspark.ml.linalg import Vectors

df = spark.createDataFrame([
    (Vectors.dense([0.0, 1.0, -2.0, 3.0]),),
    (Vectors.dense([-1.0, 2.0, 4.0, -7.0]),),
    (Vectors.dense([14.0, -2.0, -5.0, 1.0]),)], ["features"])

dct = DCT(inverse=False, inputCol="features", outputCol="featuresDCT")

dctDf = dct.transform(df)

for dcts in dctDf.select("featuresDCT").take(3):
    print(dcts)

STringindexer

算法介紹：

StringIndexer將字符串標籤編碼爲標籤指標。指標取值範圍爲[0,numLabels]，按照標籤出現頻率排序，所以出現最頻繁的標籤其指標爲0。如果輸入列爲數值型，我們先將之映射到字符串然後再對字符串的值進行指標。如果下游的管道節點需要使用字符串－指標標籤，則必須將輸入和鑽還爲字符串－指標列名。

示例：

假設我們有DataFrame數據含有id和category兩列：

id | category

----|----------

 0  | a

 1  | b

 2  | c

 3  | a

 4  | a

 5  | c

category是有3種取值的字符串列，使用StringIndexer進行轉換後我們可以得到如下輸出：

id | category |categoryIndex

----|----------|---------------

 0  |a        | 0.0

 1  |b        | 2.0

 2  |c        | 1.0

 3  |a        | 0.0

 4  |a        | 0.0

 5  |c        | 1.0

另外，如果在轉換新數據時出現了在訓練中未出現的標籤，StringIndexer將會報錯（默認值）或者跳過未出現的標籤實例。

示例調用：

Scala：

import org.apache.spark.ml.feature.StringIndexer

val df = spark.createDataFrame(
  Seq((0, "a"), (1, "b"), (2, "c"), (3, "a"), (4, "a"), (5, "c"))
).toDF("id", "category")

val indexer = new StringIndexer()
  .setInputCol("category")
  .setOutputCol("categoryIndex")

val indexed = indexer.fit(df).transform(df)
indexed.show()

Java：

import java.util.Arrays;
import java.util.List;

import org.apache.spark.ml.feature.StringIndexer;
import org.apache.spark.sql.Dataset;
import org.apache.spark.sql.Row;
import org.apache.spark.sql.RowFactory;
import org.apache.spark.sql.types.StructField;
import org.apache.spark.sql.types.StructType;

import static org.apache.spark.sql.types.DataTypes.*;

List<Row> data = Arrays.asList(
  RowFactory.create(0, "a"),
  RowFactory.create(1, "b"),
  RowFactory.create(2, "c"),
  RowFactory.create(3, "a"),
  RowFactory.create(4, "a"),
  RowFactory.create(5, "c")
);
StructType schema = new StructType(new StructField[]{
  createStructField("id", IntegerType, false),
  createStructField("category", StringType, false)
});
Dataset<Row> df = spark.createDataFrame(data, schema);
StringIndexer indexer = new StringIndexer()
  .setInputCol("category")
  .setOutputCol("categoryIndex");
Dataset<Row> indexed = indexer.fit(df).transform(df);
indexed.show();

Python：

from pyspark.ml.feature import StringIndexer

df = spark.createDataFrame(
    [(0, "a"), (1, "b"), (2, "c"), (3, "a"), (4, "a"), (5, "c")],
    ["id", "category"])
indexer = StringIndexer(inputCol="category", outputCol="categoryIndex")
indexed = indexer.fit(df).transform(df)
indexed.show()

IndexToString

算法介紹：

       與StringIndexer對應，IndexToString將指標標籤映射回原始字符串標籤。一個常用的場景是先通過StringIndexer產生指標標籤，然後使用指標標籤進行訓練，最後再對預測結果使用IndexToString來獲取其原始的標籤字符串。

示例：

假設我們有如下的DataFrame包含id和categoryIndex兩列：

id | categoryIndex

----|---------------

 0  | 0.0

 1  | 2.0

 2  | 1.0

 3  | 0.0

 4  | 0.0

 5  | 1.0

使用originalCategory我們可以獲取其原始的標籤字符串如下：

id | categoryIndex| originalCategory

----|---------------|-----------------

 0  |0.0           | a

 1  |2.0           | b

 2  |1.0           | c

 3  |0.0           | a

 4  |0.0           | a

 5  |1.0           | c

示例調用：

Scala：

import org.apache.spark.ml.feature.{IndexToString, StringIndexer}

val df = spark.createDataFrame(Seq(
  (0, "a"),
  (1, "b"),
  (2, "c"),
  (3, "a"),
  (4, "a"),
  (5, "c")
)).toDF("id", "category")

val indexer = new StringIndexer()
  .setInputCol("category")
  .setOutputCol("categoryIndex")
  .fit(df)
val indexed = indexer.transform(df)

val converter = new IndexToString()
  .setInputCol("categoryIndex")
  .setOutputCol("originalCategory")

val converted = converter.transform(indexed)
converted.select("id", "originalCategory").show()

Java：

import java.util.Arrays;
import java.util.List;

import org.apache.spark.ml.feature.IndexToString;
import org.apache.spark.ml.feature.StringIndexer;
import org.apache.spark.ml.feature.StringIndexerModel;
import org.apache.spark.sql.Row;
import org.apache.spark.sql.RowFactory;
import org.apache.spark.sql.types.DataTypes;
import org.apache.spark.sql.types.Metadata;
import org.apache.spark.sql.types.StructField;
import org.apache.spark.sql.types.StructType;

List<Row> data = Arrays.asList(
  RowFactory.create(0, "a"),
  RowFactory.create(1, "b"),
  RowFactory.create(2, "c"),
  RowFactory.create(3, "a"),
  RowFactory.create(4, "a"),
  RowFactory.create(5, "c")
);
StructType schema = new StructType(new StructField[]{
  new StructField("id", DataTypes.IntegerType, false, Metadata.empty()),
  new StructField("category", DataTypes.StringType, false, Metadata.empty())
});
Dataset<Row> df = spark.createDataFrame(data, schema);

StringIndexerModel indexer = new StringIndexer()
  .setInputCol("category")
  .setOutputCol("categoryIndex")
  .fit(df);
Dataset<Row> indexed = indexer.transform(df);

IndexToString converter = new IndexToString()
  .setInputCol("categoryIndex")
  .setOutputCol("originalCategory");
Dataset<Row> converted = converter.transform(indexed);
converted.select("id", "originalCategory").show();

Python：

from pyspark.ml.feature import IndexToString, StringIndexer

df = spark.createDataFrame(
    [(0, "a"), (1, "b"), (2, "c"), (3, "a"), (4, "a"), (5, "c")],
    ["id", "category"])

stringIndexer = StringIndexer(inputCol="category", outputCol="categoryIndex")
model = stringIndexer.fit(df)
indexed = model.transform(df)

converter = IndexToString(inputCol="categoryIndex", outputCol="originalCategory")
converted = converter.transform(indexed)

converted.select("id", "originalCategory").show()

OneHotEncoder

算法介紹：

    獨熱編碼將標籤指標映射爲二值向量，其中最多一個單值。這種編碼被用於將種類特徵使用到需要連續特徵的算法，如邏輯迴歸等。

示例調用：

Scala：

import org.apache.spark.ml.feature.{OneHotEncoder, StringIndexer}

val df = spark.createDataFrame(Seq(
  (0, "a"),
  (1, "b"),
  (2, "c"),
  (3, "a"),
  (4, "a"),
  (5, "c")
)).toDF("id", "category")

val indexer = new StringIndexer()
  .setInputCol("category")
  .setOutputCol("categoryIndex")
  .fit(df)
val indexed = indexer.transform(df)

val encoder = new OneHotEncoder()
  .setInputCol("categoryIndex")
  .setOutputCol("categoryVec")
val encoded = encoder.transform(indexed)
encoded.select("id", "categoryVec").show()

Java：

import java.util.Arrays;
import java.util.List;

import org.apache.spark.ml.feature.OneHotEncoder;
import org.apache.spark.ml.feature.StringIndexer;
import org.apache.spark.ml.feature.StringIndexerModel;
import org.apache.spark.sql.Dataset;
import org.apache.spark.sql.Row;
import org.apache.spark.sql.RowFactory;
import org.apache.spark.sql.types.DataTypes;
import org.apache.spark.sql.types.Metadata;
import org.apache.spark.sql.types.StructField;
import org.apache.spark.sql.types.StructType;

List<Row> data = Arrays.asList(
  RowFactory.create(0, "a"),
  RowFactory.create(1, "b"),
  RowFactory.create(2, "c"),
  RowFactory.create(3, "a"),
  RowFactory.create(4, "a"),
  RowFactory.create(5, "c")
);

StructType schema = new StructType(new StructField[]{
  new StructField("id", DataTypes.IntegerType, false, Metadata.empty()),
  new StructField("category", DataTypes.StringType, false, Metadata.empty())
});

Dataset<Row> df = spark.createDataFrame(data, schema);

StringIndexerModel indexer = new StringIndexer()
  .setInputCol("category")
  .setOutputCol("categoryIndex")
  .fit(df);
Dataset<Row> indexed = indexer.transform(df);

OneHotEncoder encoder = new OneHotEncoder()
  .setInputCol("categoryIndex")
  .setOutputCol("categoryVec");
Dataset<Row> encoded = encoder.transform(indexed);
encoded.select("id", "categoryVec").show();

Python：

from pyspark.ml.feature import OneHotEncoder, StringIndexer

df = spark.createDataFrame([
    (0, "a"),
    (1, "b"),
    (2, "c"),
    (3, "a"),
    (4, "a"),
    (5, "c")
], ["id", "category"])

stringIndexer = StringIndexer(inputCol="category", outputCol="categoryIndex")
model = stringIndexer.fit(df)
indexed = model.transform(df)
encoder = OneHotEncoder(dropLast=False, inputCol="categoryIndex", outputCol="categoryVec")
encoded = encoder.transform(indexed)
encoded.select("id", "categoryVec").show()

文章出處：https://blog.csdn.net/liulingyuan6/article/details/53397780?locationNum=3&fps=1