Spark ML 特征转换及处理算子实战技巧-Spark商业ML实战
本套技术专栏是作者（秦凯新）平时工作的总结和升华，通过从真实商业环境抽取案例进行总结和分享，并给出商业应用的调优建议和集群环境容量规划等内容，请持续关注本套博客。版权声明：禁止转载，欢迎学习。QQ邮箱地址：[email protected]，如有任何商业交流，可随时联系。
1 燃烧吧特征转换
1.1 Tokenization 分词器技术(RegexTokenizer)
Tokenization是将文本(例如句子)分割成单词，默认是空格分割
RegexTokenizer是基于正则表达式进行单词分割，默认打分割方式是’\s+’，
import org.apache.spark.ml.feature.{RegexTokenizer, Tokenizer}
import org.apache.spark.sql.SparkSession
import org.apache.spark.sql.functions._
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("id", "sentence")
=> 默认匹配的是空格
val tokenizer = new Tokenizer().setInputCol("sentence").setOutputCol("words")
val tokenized = tokenizer.transform(sentenceDataFrame)
val countTokens = udf { (words: Seq[String]) => words.length }
tokenized.select("sentence", "words").withColumn("tokens", countTokens(col("words"))).show(false)
+-----------------------------------+------------------------------------------+------+
|sentence |words |tokens|
+-----------------------------------+------------------------------------------+------+
|Hi I heard about Spark |[hi, i, heard, about, spark] |5 |
|I wish Java could use case classes |[i, wish, java, could, use, case, classes]|7 |
|Logistic,regression,models,are,neat|[logistic,regression,models,are,neat] |1 |
+-----------------------------------+------------------------------------------+------+
=> W 表示匹配出的是除单词之外的任何分隔符
val regexTokenizer = new
RegexTokenizer().setInputCol("sentence").setOutputCol("words").setPattern("\\W")
val regexTokenized = regexTokenizer.transform(sentenceDataFrame)
regexTokenized.select("sentence", "words").withColumn("tokens", countTokens(col("words"))).show(false)
scala> regexTokenized.select("sentence", "words").withColumn("tokens", countTokens(col("words"))).show(false)
+-----------------------------------+------------------------------------------+------+
|sentence |words |tokens|
+-----------------------------------+------------------------------------------+------+
|Hi I heard about Spark |[hi, i, heard, about, spark] |5 |
|I wish Java could use case classes |[i, wish, java, could, use, case, classes]|7 |
|Logistic,regression,models,are,neat|[logistic, regression, models, are, neat] |5 |
+-----------------------------------+------------------------------------------+------+
=> w 表示为 [a-zA-Z0-9]，setGaps(false)表示匹配的不是空格，而是直接匹配出单词
val regexTokenizer = new RegexTokenizer().setInputCol("sentence").setOutputCol("words").setPattern("\\w+").setGaps(false)
val regexTokenized = regexTokenizer.transform(sentenceDataFrame)
regexTokenized.select("sentence", "words").withColumn("tokens", countTokens(col("words"))).show(false)
regexTokenized.select("sentence", "words").withColumn("tokens", countTokens(col("words"))).show(false)
+-----------------------------------+------------------------------------------+------+
|sentence |words |tokens|
+-----------------------------------+------------------------------------------+------+
|Hi I heard about Spark |[hi, i, heard, about, spark] |5 |
|I wish Java could use case classes |[i, wish, java, could, use, case, classes]|7 |
|Logistic,regression,models,are,neat|[logistic, regression, models, are, neat] |5 |
+-----------------------------------+------------------------------------------+------+
1.2 移除停用词
发现 I ，and ,had , a 被移除
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", "balloon")),
(1, Seq("Mary", "had", "a", "little", "lamb"))
)).toDF("id", "raw")
remover.transform(dataSet).show(false)
scala> remover.transform(dataSet).show(false)
+---+----------------------------+--------------------+
|id |raw |filtered |
+---+----------------------------+--------------------+
|0 |[I, saw, the, red, balloon] |[saw, red, balloon] |
|1 |[Mary, had, a, little, lamb]|[Mary, little, lamb]|
+---+----------------------------+--------------------+
## 要求的默认停用词列表
requirement failed: US is not in the supported language list: french, spanish, german, finnish, turkish, english, russian, norwegian, dutch, danish, hungarian, italian, swedish, portuguese.
scala> StopWordsRemover.loadDefaultStopWords("english")
res17: Array[String] = Array(i, me, my, myself, we, our, ours, ourselves, you, your, yours, yourself, yourselves, he, him, his, himself, she, her, hers, herself, it, its, itself, they, them, their, theirs, themselves, what, which, who, whom, this, that, these, those, am, is, are, was, were, be, been, being, have, has, had, having, do, does, did, doing, a, an, the, and, but, if, or, because, as, until, while, of, at, by, for, with, about, against, between, into, through, during, before, after, above, below, to, from, up, down, in, out, on, off, over, under, again, further, then, once, here, there, when, where, why, how, all, any, both, each, few, more, most, other, some, such, no, nor, not, only, own, same, so, than, too, very, s, t, can, will, just, don, should, now, i'll, you'll, he'll...
1.3 n-gram (得到组合序列)
每个n-gram由n个连续字的空格分隔的字符串表示。 如果输入序列包含少于n个字符串，则不会生成输出。
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("id", "words")
val ngram = new NGram().setN(3).setInputCol("words").setOutputCol("ngrams")
val ngramDataFrame = ngram.transform(wordDataFrame)
ngramDataFrame.select("ngrams").show(false)
+--------------------------------------------------------------------------------+
|ngrams |
+--------------------------------------------------------------------------------+
|[Hi I heard, I heard about, heard about Spark] |
|[I wish Java, wish Java could, Java could use, could use case, use case classes]|
|[Logistic regression models, regression models are, models are neat] |
+--------------------------------------------------------------------------------+
1.4 二值化（还是比较厉害的）
二值化是将数字特征阈值为二进制（0/1）特征的过程。
Binarizer接受通用参数inputCol和outputCol以及二进制阈值。 大于阈值的特征值被二进制化为1.0; 等于或小于阈值的值被二值化为0.0。 inputCol支持Vector和Double类型。
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("id", "feature")
val binarizer: Binarizer = new Binarizer().setInputCol("feature").setOutputCol("binarized_feature").setThreshold(0.5)
val binarizedDataFrame = binarizer.transform(dataFrame)
scala> println(s"Binarizer output with Threshold = ${binarizer.getThreshold}")
Binarizer output with Threshold = 0.5
scala>binarizedDataFrame.show()
+---+-------+-----------------+
| id|feature|binarized_feature|
+---+-------+-----------------+
| 0| 0.1| 0.0|
| 1| 0.8| 1.0|
| 2| 0.2| 0.0|
+---+-------+-----------------+
1.5 规范化（StandardScaler）
StandardScaler处理的对象是每一列，也就是每一维特征，将特征标准化为单位标准差或是0均值，或是0均值单位标准差。
StandardScaler=(x-u)/标准差Sn
主要有两个参数可以设置：
-
withStd: 默认为真。将数据标准化到单位标准差。
-
withMean: 默认为假。是否变换为0均值。 (此种方法将产出一个稠密输出，所以不适用于稀疏输入。)
-
StandardScaler需要fit数据，获取每一维的均值和标准差，来缩放每一维特征。
StandardScaler是一个Estimator，它可以fit数据集产生一个StandardScalerModel，用来计算汇总统计。 -
然后产生的模可以用来转换向量至统一的标准差以及（或者）零均值特征。注意如果特征的标准差为零，则该特征在向量中返回的默认值为0.0。
import org.apache.spark.ml.linalg.Vectors val dataFrame = spark.createDataFrame(Seq( (0, Vectors.dense(1.0, 0.5, -1.0)), (1, Vectors.dense(2.0, 1.0, 1.0)), (2, Vectors.dense(4.0, 10.0, 2.0)) )).toDF("id", "features") dataFrame.show +---+--------------+ | id| features| +---+--------------+ | 0|[1.0,0.5,-1.0]| | 1| [2.0,1.0,1.0]| | 2|[4.0,10.0,2.0]| +---+--------------+ val scaler = new StandardScaler().setInputCol("features").setOutputCol("scaledFeatures").setWithStd(true).setWithMean(false) val scalerModel = scaler.fit(dataFrame) val scaledData = scalerModel.transform(dataFrame) scaledData.show +---+--------------+--------------------+ | id| features| scaledFeatures| +---+--------------+--------------------+ | 0|[1.0,0.5,-1.0]|[0.65465367070797...| | 1| [2.0,1.0,1.0]|[1.30930734141595...| | 2|[4.0,10.0,2.0]|[2.61861468283190...| +---+--------------+--------------------+ scala> scaledData.rdd.foreach(println) [0,[1.0,0.5,-1.0],[0.6546536707079772,0.09352195295828246,-0.6546536707079771]] [1,[2.0,1.0,1.0],[1.3093073414159544,0.18704390591656492,0.6546536707079771]] [2,[4.0,10.0,2.0],[2.618614682831909,1.8704390591656492,1.3093073414159542]] val scaler = new StandardScaler().setInputCol("features").setOutputCol("scaledFeatures").setWithStd(true).setWithMean(true) val scalerModel = scaler.fit(dataFrame) val scaledData = scalerModel.transform(dataFrame) scaledData.show scala> scaledData.rdd.foreach(println) [0,[1.0,0.5,-1.0],[-0.8728715609439697,-0.6234796863885498,-1.0910894511799618]] [1,[2.0,1.0,1.0],[-0.2182178902359925,-0.5299577334302673,0.2182178902359924]] [2,[4.0,10.0,2.0],[1.0910894511799618,1.1534374198188169,0.8728715609439696]1.6 正则化（Normalizer） --面向行
范数是一种强化了的距离概念，它在定义上比距离多了一条数乘的运算法则。有时候为了便于理解，我们可以把范数当作距离来理解。
-
L1范数是我们经常见到的一种范数，它的定义如下：
-
L2范数是我们最常见最常用的范数了，我们用的最多的度量距离欧氏距离就是一种L2范数，它的定义如下：
-
是L无穷范数，它主要被用来度量向量元素的最大值，与L0一样，通常情况下表示为：
import org.apache.spark.ml.feature.Normalizer 正则化每个向量到1阶范数 将每一行的规整为1阶范数为1的向量，1阶范数即所有值绝对值之和 val normalizer = new Normalizer().setInputCol("features") .setOutputCol("normFeatures").setP(1.0) l1NormData.show() +---+--------------+------------------+ | id| features| normFeatures| +---+--------------+------------------+ | 0|[1.0,0.5,-1.0]| [0.4,0.2,-0.4]| | 1| [2.0,1.0,1.0]| [0.5,0.25,0.25]| | 2|[4.0,10.0,2.0]|[0.25,0.625,0.125]| +---+--------------+------------------+ 正则化每个向量到无穷阶范数，向量的无穷阶范数即向量中所有值中的最大值 val lInfNormData = normalizer.transform(dataFrame, normalizer.p -> Double.PositiveInfinity) lInfNormData.show() +---+--------------+--------------+ | id| features| normFeatures| +---+--------------+--------------+ | 0|[1.0,0.5,-1.0]|[1.0,0.5,-1.0]| | 1| [2.0,1.0,1.0]| [1.0,0.5,0.5]| | 2|[4.0,10.0,2.0]| [0.4,1.0,0.2]| +---+--------------+--------------+1.7 最大最小值缩放 MinMaxScaler --面向列 (value-Emin/(Emax-Emin))*[max-min]+min
Emin最小值为每一列最小值 Emax最小值为每一列最大值 MinMaxScaler作用同样是每一列，即每一维特征。将每一维特征线性地映射到指定的区间，通常是[0, 1]。 MinMaxScaler计算数据集的汇总统计量，并产生一个MinMaxScalerModel。 注意因为零值转换后可能变为非零值，所以即便为稀疏输入，输出也可能为稠密向量。 该模型可以将独立的特征的值转换到指定的范围内。 它也有两个参数可以设置： min: 默认为0。指定区间的下限。 max: 默认为1。指定区间的上限。 import org.apache.spark.ml.feature.MinMaxScaler val dataFrame = spark.createDataFrame(Seq( (0, Vectors.dense(1.0, 0.5, -1.0)), (1, Vectors.dense(2.0, 1.0, 1.0)), (2, Vectors.dense(4.0, 10.0, 2.0)) )).toDF("id", "features") val scaler = new MinMaxScaler() .setInputCol("features").setOutputCol("scaledFeatures") // Compute summary statistics and generate MinMaxScalerModel val scalerModel = scaler.fit(dataFrame) // rescale each feature to range [min, max]. val scaledData = scalerModel.transform(dataFrame) scaledData.select("features", "scaledFeatures").show 每维特征线性地映射，最小值映射到0，最大值映射到1。 +--------------+-----------------------------------------------------------+ |features |scaledFeatures | +--------------+-----------------------------------------------------------+ |[1.0,0.5,-1.0]|[0.0,0.0,0.0] | |[2.0,1.0,1.0] |[0.3333333333333333,0.05263157894736842,0.6666666666666666]| |[4.0,10.0,2.0]|[1.0,1.0,1.0] | +--------------+-----------------------------------------------------------+1.8 最大值-绝对值缩放MaxAbsScaler（面向列-value除以绝对值最大值）
MaxAbsScaler将每一维的特征变换到[-1,1]闭区间上，通过除以每一维特征上的最大的绝对值，它不会平移整个分布，也不会破坏原来每一个特征向量的稀疏性。因为它不会转移／集中数据，所以不会破坏数据的稀疏性。
import org.apache.spark.ml.feature.MaxAbsScaler val scaler = new MaxAbsScaler() .setInputCol("features") .setOutputCol("scaledFeatures") val scalerModel = scaler.fit(dataFrame) 1] val scaledData = scalerModel.transform(dataFrame) scaledData.select("features", "scaledFeatures").show() // 每一维的绝对值的最大值为[4, 10, 2] +--------------+----------------+ | features| scaledFeatures| +--------------+----------------+ |[1.0,0.5,-1.0]|[0.25,0.05,-0.5]| | [2.0,1.0,1.0]| [0.5,0.1,0.5]| |[4.0,10.0,2.0]| [1.0,1.0,1.0]| +--------------+----------------+
1.9 独热编码（OneHotEncoderEstimator）- 面向列
把每一列的所有可能编码成向量形式：如： 0被编码为：（1,0） 也即一个值变成一个向量。
import org.apache.spark.ml.feature.OneHotEncoderEstimator
val df = spark.createDataFrame(Seq(
(0.0, 1.0),
(1.0, 0.0),
(2.0, 1.0),
(0.0, 2.0),
(0.0, 1.0),
(2.0, 0.0)
)).toDF("categoryIndex1", "categoryIndex2")
val encoder = new OneHotEncoderEstimator().setInputCols(Array("categoryIndex1", "categoryIndex2")) .setOutputCols(Array("categoryVec1", "categoryVec2"))
val model = encoder.fit(df)
val encoded = model.transform(df)
encoded.show()
最终结果：
+--------------+--------------+-------------+-------------+
|categoryIndex1|categoryIndex2| categoryVec1| categoryVec2|
+--------------+--------------+-------------+-------------+
| 0.0| 1.0|(2,[0],[1.0])|(2,[1],[1.0])|
| 1.0| 0.0|(2,[1],[1.0])|(2,[0],[1.0])|
| 2.0| 1.0| (2,[],[])|(2,[1],[1.0])|
| 0.0| 2.0|(2,[0],[1.0])| (2,[],[])|
| 0.0| 1.0|(2,[0],[1.0])|(2,[1],[1.0])|
| 2.0| 0.0| (2,[],[])|(2,[0],[1.0])|
+--------------+--------------+-------------+-------------+
把每一列的所有可能编码成向量形式：如：第一行0被编码为：（1,0,0）
val encoder = new OneHotEncoderEstimator().setInputCols(Array("categoryIndex1", "categoryIndex2")) .setOutputCols(Array("categoryVec1", "categoryVec2")).setDropLast(false)
val model = encoder.fit(df)
val encoded = model.transform(df)
encoded.show()
+--------------+--------------+-------------+-------------+
|categoryIndex1|categoryIndex2| categoryVec1| categoryVec2|
+--------------+--------------+-------------+-------------+
| 0.0| 1.0|(3,[0],[1.0])|(3,[1],[1.0])|
| 1.0| 0.0|(3,[1],[1.0])|(3,[0],[1.0])|
| 2.0| 1.0|(3,[2],[1.0])|(3,[1],[1.0])|
| 0.0| 2.0|(3,[0],[1.0])|(3,[2],[1.0])|
| 0.0| 1.0|(3,[0],[1.0])|(3,[1],[1.0])|
| 2.0| 0.0|(3,[2],[1.0])|(3,[0],[1.0])|
+--------------+--------------+-------------+-------------+
1.9 字符串-索引变换
根据字符串出现的频率来定义标签列，出现频率最高者为0，因此可以看到0,1,2…,但是若出现训练的模型中没有测试的值，将会报错，此时需要setHandleInvalid，跳过即可。
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 model =indexer.fit(df)
val indexed = model.transform(df)
indexed.show()
scala> indexed.show
+---+--------+-------------+
| 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|
+---+--------+-------------+
训练模型中没有d时，会报错，所以需要设置setHandleInvalid
val df_test = spark.createDataFrame(
Seq((0, "a"), (1, "b"), (2, "c"), (3, "a"), (4, "a"), (5, "c"),(5, "d"))
).toDF("id", "category")
val indexer_test = new StringIndexer().setInputCol("category").setOutputCol("categoryIndex").setHandleInvalid("skip")
val model =indexer_test.fit(df)
val indexed_test = model.transform(df_test)
结语
今天是我的CSDN技术专栏突破50篇的日子，值得庆祝！
秦凯新 于深圳 2018 11 18 1:47