我是一只会飞的羊
我是一只会飞的羊

sparkMlib_doc_1.0

模型输入输出对应关系

  1. 输入表（hive）——模型参数——输出模型（hdfs）
    • DecisionTree
    • GBTC
    • LogisticRegression
    • NaiveBayes
    • RandomForest
    • BisectingKMeans
    • IDFTrain
    • ALS
    • DecisionTreeRegression
    • LinearRegression
    • RandomForestRegression

示例：
'{"InputHiveTable":"DecisionTree"}' '{"LabelCol":"label","FeaturesCol":"features","PredictionCol":"prediction","MaxIter":10}'
'{"OutputModelTarget":"hdfs:///user/model/GBTC","Overwrite":true}'

  1. 输入表（hive）——输入模型（hdfs）——输出表（hive）
    • DecisionTreePre
    • GBTCPre
    • LogisticRegressionPre
    • NaiveBayesPre
    • RandomForestPre
    • BisectingKMeansPre
    • TFIDFVectorize
    • ALSPre
    • DecisionTreeRegressionPre
    • LinearRegressionPre
    • RandomForestRegressionPre

示例：
'{"InputHiveTable":"DecisionTree"}'
'{"InputModelSource":"hdfs:///user/model/DecisionTreeModel"}'
'{"OutputHiveTable":"DecisionTreeTestPre"}'

  1. 输入表（hive）——模型参数——输出表（hive）
    • Binarizer
    • Bucketizer
    • ChiSqSelector
    • CountVectorzer
    • DCT
    • TFVectorize
    • IndexToString
    • MinMaxScaler
    • NGram
    • Normalizer
    • OneHotEncoder
    • PCA
    • RegexTokenizer
    • StandardScaler
    • StringIndexer
    • VectorAssembler
    • VectorIndexer

示例
'{"InputHiveTable":"Binarizer"}'
'{"InputCol":"feature","OutputCol":"binarized_feature","Threshold":0.5}
'{"OutputHiveTable":"Binarizer"}'

  1. 模型名称——输入表（hive）——输入模型（hdfs）——
    模型参数——输出模型（hdfs）
    • BinaryClassificationMetrics
    • RegressionMetrics

示例
'DecisionTreeRegression'
'{"InputHiveTable":"DecisionTree"}'
'{"InputModelSource":"hdfs:///user/model/DecisionTreeModel"}'
'{"LabelCol":"label","PredictionCol":"prediction"}'
'{"OutPutJsonTarget":"hdfs:///user/example/jsonss"}'

  1. 输入表（hive）——模型参数——输出结果（json格式保存到hdfs）
    • StatisticalSummary
    • Correlations

'{"InputHiveTable":"DecisionTree"}'
'{"InputCols":["f1","f2","f3","f4"],"ColLabels":["f1","f2","f3","f4"]}'
'{"OutPutJsonTarget":"hdfs:///user/example/jsonss"}'

  1. 输入表（hive）——模型参数
    • RandomSplit
      备注：训练集和测试集合默认保存到hive中

示例
'{"InputHiveTable":"DecisionTreeRegression"}'
'{"RandomRate":0.5}'
'{"OutputTrainingSetTable":"trainSet"}'
'{"OutputTestSetTable":"testSet"}'

特征

Feature特征提取

  1. TFVectorize
    1.1 参数
参数名称 参数类型 参数描述 默认值 是否必选
InputCol string Param for input column name. null true
OutputCol string Param for output column name. output true
NumFeatures int Number of features. Should be greater than 0. 20 false
Binary bool Binary toggle to control term frequency counts. false false

1.2 测试用例
feature.TFVectorize
1.3 提交参数
spark-submit --class grimoire.ml.feature.conjure.TFVectorize --master local[4] 'grimoire-assembly-0.1.0.jar' '{"InputHiveTable":"wfwfwfwf"}' '{"InputCol":"segmented","OutputCol":"tf","NumFeatures":100,"Binary":true}' '{"OutputHiveTable":"test2"}'

  1. IDFTrainSpell
    2.1 参数
参数名称 参数类型 参数描述 默认值 是否必选
InputCol string Param for input column name. null true
MinDocFreq int The minimum number of documents in which a term should appear. 0 false

2.2 测试用例
feature.IDFTrain
2.3 提交参数示例
spark-submit --class grimoire.ml.feature.conjure.IDFTrain --master local[4] 'grimoire-assembly-0.1.0.jar' '{"InputHiveTable":"test2"}' '{"InputCol":"tf", "OutputCol":"tfidf"}' '{"OutputModelTarget":"hdfs:///user/model/idf2","Overwrite":true}'

  1. TFIDFVectorize
    3.1 参数
参数名称 参数类型 参数描述 默认值 是否必选
OutputCol string Param for output column name. output true

3.2 测试用例
feature.TFIDFVectorize
3.3 提交参数示例
spark-submit --class grimoire.ml.feature.conjure.TFIDFVectorize --master local[4] 'grimoire-assembly-0.1.0.jar' '{"InputHiveTable":"test2"}' '{"InputModelSource":"hdfs:///user/model/idf2"}' '{"OutputCol":"tfidf"}' '{"OutputHiveTable":"TFIDFVectorize"}'

  1. CountVectorizer
    4.1 参数
参数名称 参数类型 参数描述 默认值 是否必选
InputCol string Param for input column name. null true
OutputCol string Param for output column name. output true
VocabSize int Max size of the vocabulary. 262144 false
MinDF double Specifies the minimum number of different documents a term must appear in to be included in the vocabulary. If this is an integer greater than or equal to 1, this specifies the number of documents the term must appear in; if this is a double in [0,1), then this specifies the fraction of documents. 1 false
Binary boolean Binary toggle to control term frequency counts. If true, all non-zero counts are set to 1. This is useful for discrete probabilistic models that model binary events rather than integer counts. false false

4.2 测试用例
feature.CountVectorizer
4.3 提交参数
spark-submit --class grimoire.ml.feature.conjure.CountVectorizer --master local[4] 'grimoire-assembly-0.1.0.jar' '{"InputHiveTable":"countvectorizer"}' '{"InputCol":"words","OutputCol":"features","VocabSize":3,"MinDF":2}' '{"OutputHiveTable":"CountVectorzerConjure"}'

特征转换

  1. RegexTokenizer
    1.1 参数
参数名称 参数类型 参数描述 默认值 是否必选
InputCol string Param for input column name. null true
OutputCol string Param for output column name. output true
Pattern String Regex pattern used to match delimiters if gaps is true or tokens if gaps is false. \s+ false
MinTokenLength int Minimum token length, greater than or equal to 0, to avoid returning empty strings 1 false
ToLowercase boolean Indicates whether to convert all characters to lowercase before tokenizing. true false

1.2 测试用例
feature.RegexTokenizer
1.3 提交参数
spark-submit --class grimoire.ml.feature.conjure.RegexTokenizer --master local[4] 'grimoire-assembly-0.1.0.jar' '{"InputHiveTable":"RegexTokenizer"}' '{"InputCol":"sentence","OutputCol":"words","Pattern":"\\W"}' '{"OutputHiveTable":"RegexTokenizerConjure"}'

  1. NGram
    2.1 参数
参数名称 参数类型 参数描述 默认值 是否必选
InputCol string Param for input column name. null true
OutputCol string Param for output column name. output true
N int Minimum n-gram length, greater than or equal to 1. 2 true

2.2 示例
feature.NGram
2.3 提交参数
spark-submit --class grimoire.ml.feature.conjure.NGram --master local[4] 'grimoire-assembly-0.1.0.jar' '{"InputHiveTable":"NGram"}' '{"InputCol":"words","OutputCol":"ngrams","N":2}' '{"OutputHiveTable":"NGramConjure"}'

  1. Binarizer
    3.1 参数
参数名称 参数类型 参数描述 默认值 是否可选
InputCol string Param for input column name. null true
OutputCol string Param for output column name. output true
Threshold Double Param for threshold used to binarize continuous features. 0.5 false

3.2 示例
feature.Binarizer
3.3 提交参数
spark-submit --class grimoire.ml.feature.conjure.Binarizer --master local[4] 'grimoire-assembly-0.1.0.jar' '{"InputHiveTable":"Binarizer"}' '{"InputCol":"feature","OutputCol":"binarized_feature","Threshold":0.5}' '{"OutputHiveTable":"BinarizerConjure"}'

  1. PCA
    4.1 参数
参数名称 参数类型 参数描述 默认值 是否可选
InputCol string Param for input column name. null true
OutputCol string Param for output column name. output true
K Int The number of clusters to infer. Must be > 1. 4 true

4.2 测试用例
feature.PCA
4.3 提交参数
spark-submit --class grimoire.ml.feature.conjure.PCA --master local[4] 'grimoire-assembly-0.1.0.jar' '{"InputHiveTable":"PCA"}' '{"InputCol":"features","OutputCol":"pcaFeatures","K":3}' '{"OutputHiveTable":"PCAConjure"}'

  1. DCT
    5.1 参数
参数名称 参数类型 参数描述 默认值 是否可选
InputCol string Param for input column name. null true
OutputCol string Param for output column name. output true
Inverse Boolean Indicates whether to perform the inverse DCT (true) or forward DCT . null true

5.2 测试用例
feature.DCT
5.3 提交参数
spark-submit --class grimoire.ml.feature.conjure.DCT --master local[4] 'grimoire-assembly-0.1.0.jar' '{"InputHiveTable":"DCT"}' '{"InputCol":"features","OutputCol":"featuresDCT","Inverse":false}' '{"OutputHiveTable":"DCTConjure"}'

  1. StringIndexer
    6.1 参数
参数名称 参数类型 参数描述 默认值 是否可选
InputCol string Param for input column name. null true
OutputCol string Param for output column name. output true
HandleInvalid string Param for how to handle invalid entries. Options are skip (which will filter out rows with bad values), or error (which will throw an error). More options may be added later. error false

6.2 测试用例
feature.StringIndexer
6.3 提交参数
spark-submit --class grimoire.ml.feature.conjure.StringIndexer --master local[4] 'grimoire-assembly-0.1.0.jar' '{"InputHiveTable":"StringIndexer"}' '{"InputCol":"category","OutputCol":"categoryIndex"}' '{"OutputHiveTable":"StringIndexerConjure"}'

  1. IndexToString
参数名称 参数类型 参数描述 默认值 是否可选
InputCol string Param for input column name. null true
OutputCol string Param for output column name. output true

7.2 测试用例
feature.IndexToString
7.3 提交参数
spark-submit --class grimoire.ml.feature.conjure.IndexToString --master local[4] 'grimoire-assembly-0.1.0.jar' '{"InputHiveTable":"IndexToString"}' '{"InputCol":"categoryIndex","OutputCol":"originalCategory"}' '{"OutputHiveTable":"IndexToStringConjure"}'

8 OneHotEncoder
8.1 参数

参数名称 参数类型 参数描述 默认值 是否可选
InputCol string Param for input column name. null true
OutputCol string Param for output column name. output true
DropLast Boolean Whether to drop the last category in the encoded vector true false

8.2 测试用例
feature.OneHotEncoder
8.3 提交参数
spark-submit --class grimoire.ml.feature.conjure.OneHotEncoder --master local[4] 'grimoire-assembly-0.1.0.jar' '{"InputHiveTable":"OneHotEncoder"}' '{"InputCol":"categoryIndex","OutputCol":"categoryVec","DropLast":false}' '{"OutputHiveTable":"OneHotEncoderConjure"}'

  1. VectorAssembler
    9.1 参数
参数名称 参数类型 参数描述 默认值 是否可选
InputCols seq[string] Param for input column name. null true
OutputCol string Param for output column name. output true

9.2 测试用例
feature.VectorAssembler
9.3 提交参数
spark-submit --class grimoire.ml.feature.conjure.VectorAssembler --master local[4] 'grimoire-assembly-0.1.0.jar' '{"InputHiveTable":"VectorAssembler"}' '{"InputCols":["hour", "mobile", "userFeatures"],"OutputCol":"features"}' '{"OutputHiveTable":"VectorAssemblerConjure"}'

  1. Normalizer
    10.1 参数
参数名称 参数类型 参数描述 默认值 是否可选
InputCol string Param for input column name. null true
OutputCol string Param for output column name. output true
P doube Normalization in Lp space. null true

10.2 测试用例
feature.Normalizer
10.3 提交参数
spark-submit --class grimoire.ml.feature.conjure.Normalizer --master local[4] 'grimoire-assembly-0.1.0.jar' '{"InputHiveTable":"Normalizer"}' '{"InputCol":"features","OutputCol":"normFeatures","P":1}' '{"OutputHiveTable":"NormalizerConjure"}'

  1. StandardScaler
    11.1参数
参数名称 参数类型 参数描述 默认值 是否可选
InputCol string Param for input column name. null true
OutputCol string Param for output column name. output true
WithStd bool Whether to scale the data to unit standard deviation. false false
WithMean bool Whether to center the data with mean before scaling. true false

11.2 测试用例
feature.StandardScaler
11.3 提交参数
spark-submit --class grimoire.ml.feature.conjure.StandardScaler --master local[4] 'grimoire-assembly-0.1.0.jar' '{"InputHiveTable":"StandardScaler"}' '{"InputCol":"features","OutputCol":"scaledFeatures","WithStd":true}' '{"OutputHiveTable":"StandardScalerConjure"}'

  1. MinMaxScaler
    12.1 参数配置
参数名称 参数类型 参数描述 默认值 是否可选
InputCol string Param for input column name. null true
OutputCol string Param for output column name. output true

12.2 测试用例
feature.MinMaxScaler
12.3 提交参数
spark-submit --class grimoire.ml.feature.conjure.MinMaxScaler --master local[4] 'grimoire-assembly-0.1.0.jar' '{"InputHiveTable":"MinMaxScaler"}' '{"InputCol":"features","OutputCol":"scaledFeatures"}' '{"OutputHiveTable":"MinMaxScalerConjure"}'

13 Bucketizer
13.1 参数配置

参数名称 参数类型 参数描述 默认值 是否可选
InputCol string Param for input column name. null true
OutputCol string Param for output column name. output true
Splits seq[double] Parameter for mapping continuous features into buckets. output true

13.2 测试用例
feature.Bucketizer
13.3 提交参数
spark-submit --class grimoire.ml.feature.conjure.Bucketizer --master local[4] 'grimoire-assembly-0.1.0.jar' '{"InputHiveTable":"Bucketizer"}' '{"InputCol":"features","OutputCol":"bucketedFeatures","Splits":[-1000, -0.5, 0.0, 0.5, 1000]}' '{"OutputHiveTable":"BucketizerConjure"}'

  1. RandomSplit
    14.1 参数配置
参数名称 参数类型 参数描述 默认值 是否可选
RandomRate double trainset random rate. null true
Trainset string split dataset to trainset. trainset true
TestSet string split dataset to testset . testset true

13.2 测试用例
feature.RandomSplit
13.3 提交参数
spark-submit --class grimoire.ml.feature.conjure.RandomSplit --master local[4] 'grimoire-assembly-0.1.0.jar' '{"InputHiveTable":"DecisionTreeRegression"}' '{"RandomRate":0.5}'
'{"OutputTrainingSetTable":"trainSet"}'
'{"OutputTestSetTable":"testSet"}'

Feature特征选择

  1. ChiSqSelector
参数名称 参数类型 参数描述 默认值 是否必选
FeaturesCol string Column of features. features true
OutputCol string Param for output column name. output true
NumTopFeatures Int Number of features that selector will select, ordered by ascending p-value. 50 true
LabelCol String Column name of label. label true

1.2 测试用例
feature.ChiSqSelector
1.3 提交参数
spark-submit --class grimoire.ml.feature.conjure.ChiSqSelector --master local[4] 'grimoire-assembly-0.1.0.jar' '{"InputHiveTable":"ChiSqSelector"}' '{"NumTopFeatures":1,"FeaturesCol":"features","LabelCol":"clicked","OutputCol":"chisq"}' '{"OutputHiveTable":"ChiSqSelectorConjure"}'

分类

  1. LogisticRegression
参数名称 参数类型 参数描述 默认值 是否可选
MaxIter Int Max iteration of train. 100 false
RegParam double Regularize parameter. 0.0 false
ElasticNetParam double The param of α. 0.0 true
Family String binomial logistic regression or multinomial logistic regression . auto false
LabelCol String Column name of label. label true
FeaturesCol String Column of features. features true
FitIntercept boolean Param for whether to fit an intercept term true false
Standardization boolean Param for whether to standardize the training features before fitting the model. true false
Threshold double Limit of calculation 0.5 false
Tol double Param for the convergence tolerance for iterative algorithms (>= 0) 1.0E-6 false
ProbabilityCol String Param for Column name for predicted class conditional probabilities. probability false
RawPredictionCol String aram for prediction column name. RawPredictionCol false
predictionCol String Param for prediction column name. prediction true

1.2 测试用例
classify.LogisticRegression
1.3 提交参数
spark-submit --class grimoire.ml.classify.conjure.LogisticRegression --master local[4] 'grimoire-assembly-0.1.0.jar' '{"InputHiveTable":"logistregression_train"}' '{"MaxIter":10,"RegParam":0.3,"ElasticNetParam":0.8,"LabelCol":"label","FeaturesCol":"features","PredictionCol":"predic"}' '{"OutputModelTarget":"hdfs:///user/model/LogisticRegression","Overwrite":true}'

  1. LogisticRegressionPre
    2.1 测试用例
    classify.LogisticRegressionPre
    2.2 提交参数
    spark-submit --class grimoire.ml.classify.conjure.LogisticRegressionPre --master local[4] 'grimoire-assembly-0.1.0.jar' '{"InputHiveTable":"logistregression_train"}' '{"InputModelSource":"hdfs:///user/model/LogisticRegression"}' '{"OutputHiveTable":"LogisticRegressionTestPre"}'

  2. DecisionTree
    3.1 参数

参数名称 参数类型 参数描述 默认值 是否必选
LabelCol String Column name of label. label true
FeaturesCol String Column of features. features true
RawPredictionCol String Param for prediction column name. rawPrediction false
Seed long Random seed. 159147643 false
CheckpointInterval int Param for set checkpoint interval (>= 1) or disable checkpoint (-1). 10 false
Impurity String Criterion used for information gain calculation (case-insensitive). gini false
MaxBins int Maximum number of bins used for discretizing continuous features and for choosing how to split on features at each node. 32 false
MaxDepth int Maximum depth of the tree (>= 0). true false
Threshold double Limit of calculation 0.5 false
MinInfoGain double Minimum information gain for a split to be considered at a tree node. 0.0 false
MinInstancesPerNode int Minimum number of instances each child must have after split. 1 false
PredictionCol String Param for prediction column name. prediction true
ProbabilityCol String Param for Column name for predicted class conditional probabilities. probabilities false

3.2 测试用例
classify.DecisionTree
3.3 提交参数
spark-submit --class grimoire.ml.classify.conjure.DecisionTree --master local[4] 'grimoire-assembly-0.1.0.jar' '{"InputHiveTable":"DecisionTree"}' '{"LabelCol":"label","FeaturesCol":"features","PredictionCol":"prediction"}' '{"OutputModelTarget":"hdfs:///user/model/DecisionTree","Overwrite":true}'

  1. DecisionTreePre
    4.1 测试用例
    classify.DecisionTreePre
    4.2 提交参数
    spark-submit --class grimoire.ml.classify.conjure.DecisionTreePre --master local[4] 'grimoire-assembly-0.1.0.jar' '{"InputHiveTable":"DecisionTree"}' '{"InputModelSource":"hdfs:///user/model/DecisionTree"}' '{"OutputHiveTable":"DecisionTreeTestPre"}'

  2. GBTC
    5.1. 模型参数

参数名称 参数类型 参数描述 默认值 是否必选
LabelCol String Column name of label. label True
FeaturesCol String Column of features. features True
MaxIter int Max iteration of train (default: 20) 20 False
Impurity String Criterion used for information gain calculation (case-insensitive). (default: gini) gini False
MaxBins int Maximum number of bins used for discretizing continuous features and for choosing how to split on features at each node. (default: 32) 32 False
MaxDepth int Maximum depth of the tree (>= 0). (default: 5) 5 False
MinInfoGain double Minimum information gain for a split to be considered at a tree node. (default: 0.0) 0.0 False
CheckpointInterval int Param for set checkpoint interval (>= 1) or disable checkpoint (-1). (default: 10) 10 False
MinInstancesPerNode int Minimum number of instances each child must have after split. (default: 1) 1 False
PredictionCol String Param for prediction column name. prediction True
Seed long Random seed. (default: -1287390502) -1287390502 False
RawPredictionCol String Param for prediction column name. (default: rawPrediction) rawPrediction False
subsamplingRate double Fraction of the training data used for learning each decision tree, in range (0, 1]. (default: 1.0) 1.0 False
LossType String Loss function which GBT tries to minimize. (default: squared) squared False
StepSize double Set the initial step size of SGD for the first step. Default 1.0. In subsequent steps, the step size will decrease with stepSize/sqrt(t) (default: 0.1) 0.1 False

5.2 测试用例
classify.GBTC
5.3 提交参数
spark-submit --class grimoire.ml.classify.conjure.GBTC --master local[4] 'grimoire-assembly-0.1.0.jar' '{"InputHiveTable":"DecisionTree"}' '{"LabelCol":"label","FeaturesCol":"features","PredictionCol":"prediction","MaxIter":10,"LossType":"logistic"}' '{"OutputModelTarget":"hdfs:///user/model/GBTC","Overwrite":true}'

  1. GBTCPre
    6.1 测试用例
    classify.GBTCPre
    6.2 提交参数
    spark-submit --class grimoire.ml.classify.conjure.GBTCPre --master local[4] 'grimoire-assembly-0.1.0.jar' '{"InputHiveTable":"DecisionTree"}' '{"InputModelSource":"hdfs:///user/model/GBTC"}' '{"OutputHiveTable":"GBTCTestPre"}'

  2. NaiveBayes
    7.1 模型参数

参数名称 参数类型 参数描述 默认值 是否必选
LabelCol String Column name of label. label True
FeaturesCol String Column of features. features True
WeightCol String Param for weight column name. (default: null) None False
ModelType String The model type which is a string (case-sensitive). (default: multinomial) multinomial False
smothing double The smoothing parameter. (default: 1.0) 1.0 False
PredictionCol String Param for prediction column name. prediction True
ProbabilityCol String Param for Column name for predicted class conditional probabilities. (default: probabilities) probabilities False
RawPredictionCol String Param for prediction column name. (default: rawPrediction) rawPrediction False
Thresholds Seq Param for Thresholds in multi-class classification to adjust the probability of predicting each class. (default: null) None False

7.2 测试用例
classify.NaiveBayes
7.3 提交参数
spark-submit --class grimoire.ml.classify.conjure.NaiveBayes --master local[4] 'grimoire-assembly-0.1.0.jar' '{"InputHiveTable":"DecisionTree"}' '{"LabelCol":"label","FeaturesCol":"features","PredictionCol":"prediction"}' '{"OutputModelTarget":"hdfs:///user/model/NaiveBayes","Overwrite":true}'

  1. NaiveBayesPre
    8.1 测试用例
    classify.NaiveBayesPre
    8.2 提交参数
    spark-submit --class grimoire.ml.classify.conjure.NaiveBayesPre --master local[4] 'grimoire-assembly-0.1.0.jar' '{"InputHiveTable":"DecisionTree"}' '{"InputModelSource":"hdfs:///user/model/NaiveBayes"}' '{"OutputHiveTable":"NaiveBayesTestPre"}'

  2. RandomForest
    9.1 模型参数

参数名称 参数类型 参数描述 默认值 是否必选
LabelCol String Column name of label. label True
FeaturesCol String Column of features. features True
NumTrees int Number of trees to train. 20 True
Impurity String Criterion used for information gain calculation (case-insensitive). (default: gini) gini False
MaxBins int Maximum number of bins used for discretizing continuous features and for choosing how to split on features at each node. (default: 32) 32 False
MaxDepth int Maximum depth of the tree (>= 0). (default: 5) 5 False
MinInfoGain double Minimum information gain for a split to be considered at a tree node. (default: 0.0) 0.0 False
CheckpointInterval int Param for set checkpoint interval (>= 1) or disable checkpoint (-1). (default: 10) 10 False
MinInstancesPerNode int Minimum number of instances each child must have after split. (default: 1) 1 False
PredictionCol String Param for prediction column name. prediction True
FeatureSubsetStrategy String The number of features to consider for splits at each tree node. (default: auto) auto False
ProbabilityCol String Param for Column name for predicted class conditional probabilities. (default: probabilities) probabilities False
Thresholds Seq Param for Thresholds in multi-class classification to adjust the probability of predicting each class. (default: null) None False
Seed long Random seed. (default: 159147643) 159147643 False
RawPredictionCol String Param for prediction column name. (default: rawPrediction) rawPrediction False
subsamplingRate double Fraction of the training data used for learning each decision tree, in range (0, 1]. (default: 1.0) 1.0 False

10.2 测试用例
classify.RandomForest
10.3 提交参数
spark-submit --class grimoire.ml.classify.conjure.RandomForest --master local[4] 'grimoire-assembly-0.1.0.jar' '{"InputHiveTable":"DecisionTree"}' '{"LabelCol":"label","FeaturesCol":"features","PredictionCol":"prediction","NumTrees":10}' '{"OutputModelTarget":"hdfs:///user/model/RandomForest","Overwrite":true}'

  1. RandomForestPre
    8.1 测试用例
    classify.RandomForestPre
    8.2 提交参数
    spark-submit --class grimoire.ml.classify.conjure.RandomForestPre --master local[4] 'grimoire-assembly-0.1.0.jar' '{"InputHiveTable":"DecisionTree"}' '{"InputModelSource":"hdfs:///user/model/RandomForest"}' '{"OutputHiveTable":"RandomForestTestPre"}'

回归

  1. DecisionTreeRegression
    1.1 模型参数
参数名称 参数类型 参数描述 默认值 是否必选
LabelCol String Column name of label. label True
FeaturesCol String Column of features. features True
PredictionCol String Param for prediction column name. prediction True
varianceCol String Param for Column name for the biased sample variance of prediction. (default: null) None False
CacheNodeIds boolean (default: false) false False
CheckpointInterval int Param for set checkpoint interval (>= 1) or disable checkpoint (-1). (default: 10) 10 False
Impurity String Criterion used for information gain calculation (case-insensitive). (default: variance) variance False
MaxBins int Maximum number of bins used for discretizing continuous features and for choosing how to split on features at each node. (default: 32) 32 False
MaxDepth int Maximum depth of the tree (>= 0). (default: 5) 5 False
maxMemoryInMB int Max number of memory intput. (default: 256) 256 False
MinInfoGain double Minimum information gain for a split to be considered at a tree node. (default: 0.0) 0.0 False
MinInstancesPerNode int Minimum number of instances each child must have after split. (default: 1) 1 False
Seed long Random seed. (default: 926680331) 926680331 False

1.2 测试用例
regression.DecisionTreeRegression
1.3 提交参数
spark-submit --class grimoire.ml.regression.conjure.DecisionTreeRegression --master local[4] 'grimoire-assembly-0.1.0.jar' '{"InputHiveTable":"DecisionTreeRegression"}' '{"LabelCol":"label","FeaturesCol":"features","PredictionCol":"prediction"}' '{"OutputModelTarget":"hdfs:///user/model/DecisionTreeRegression","Overwrite":true}'

DecisionTreeRegressionPre
2.1 测试用例
regression.RandomForestPre
2.2 提交参数
spark-submit --class grimoire.ml.regression.conjure.DecisionTreeRegressionPre --master local[4] 'grimoire-assembly-0.1.0.jar' '{"InputHiveTable":"DecisionTreeRegression"}' '{"InputModelSource":"hdfs:///user/model/DecisionTreeRegression"}' '{"OutputHiveTable":"DecisionTreeRegressionPre"}'

3 LinearRegression

参数名称 参数类型 参数描述 默认值 是否必选
MaxIter int Max iteration of train (default: 100) 100 False
RegParam double Regularize parameter (default: 0.0) 0.0 False
ElasticNetParam double The param of α. (default: 0.0) 0.0 False
FitIntercept boolean Param for whether to fit an intercept term. (default: true) true False
LabelCol String Column name of label. label True
FeaturesCol String Column of features. features True
WeightCol String Param for weight column name. (default: null) None False
AggregationDepth int the depth of aggregation (default: 2) 2 False
Standardization boolean Param for whether to standardize the training features before fitting the model. (default: true) true False
Solver String Param for the solver algorithm for optimization. (default: auto) auto False
Tol double Param for the convergence tolerance for iterative algorithms (>= 0). (default: 1.0E-6) 1.0E-6 False
PredictionCol String Param for prediction column name. prediction True

3.1 测试用例
regression.LinearRegression
3.2 提交参数
spark-submit --class grimoire.ml.regression.conjure.LinearRegression --master local[4] 'grimoire-assembly-0.1.0.jar' '{"InputHiveTable":"DecisionTreeRegression"}' '{"LabelCol":"label","FeaturesCol":"features","PredictionCol":"prediction","MaxIter":10,"RegParam":0.3,"ElasticNetParam":0.8}' '{"OutputModelTarget":"hdfs:///user/model/LinearRegression","Overwrite":true}'

  1. LinearRegressionPre
    4.1 测试用例
    regression.LinearRegressionPre
    4.2 提交参数
    spark-submit --class grimoire.ml.regression.conjure.LinearRegressionPre --master local[4] 'grimoire-assembly-0.1.0.jar' '{"InputHiveTable":"DecisionTreeRegression"}' '{"InputModelSource":"hdfs:///user/model/LinearRegression"}' '{"OutputHiveTable":"LinearRegressionPre"}'

5 RandomForestRegression

参数名称 参数类型 参数描述 默认值 是否必选
LabelCol String Column name of label. label True
FeaturesCol String Column of features. features True
PredictionCol String Param for prediction column name. prediction True
CacheNodeIds boolean (default: false) false False
CheckpointInterval int Param for set checkpoint interval (>= 1) or disable checkpoint (-1). (default: 10) 10 False
Seed long Random seed. (default: 235498149) 235498149 False
Impurity String Criterion used for information gain calculation (case-insensitive). (default: variance) variance False
MaxBins int Maximum number of bins used for discretizing continuous features and for choosing how to split on features at each node. (default: 32) 32 False
MaxDepth int Maximum depth of the tree (>= 0). (default: 5) 5 False
maxMemoryInMB int Max number of memory intput. (default: 256) 256 False
MinInfoGain double Minimum information gain for a split to be considered at a tree node. (default: 0.0) 0.0 False
MinInstancesPerNode int Minimum number of instances each child must have after split. (default: 1) 1 False
NumTrees int Number of trees to train. 10 True
subsamplingRate double Fraction of the training data used for learning each decision tree, in range (0, 1]. (default: 1.0) 1.0 False
FeatureSubsetStrategy String The number of features to consider for splits at each tree node. (default: auto) auto False

5.1 测试用例
regression.RandomForestRegression
5.2 提交参数
spark-submit --class grimoire.ml.regression.conjure.RandomForestRegression --master local[4] 'grimoire-assembly-0.1.0.jar' '{"InputHiveTable":"DecisionTreeRegression"}' '{"LabelCol":"label","FeaturesCol":"features","PredictionCol":"prediction","NumTrees":20}' '{"OutputModelTarget":"hdfs:///user/model/RandomForestRegression","Overwrite":true}'

  1. RandomForestRegressionPre
    6.1 测试用例
    regression.RandomForestRegressionPre
    6.2 提交参数
    spark-submit --class grimoire.ml.regression.conjure.RandomForestRegressionPre --master local[4] 'grimoire-assembly-0.1.0.jar' '{"InputHiveTable":"DecisionTreeRegression"}' '{"InputModelSource":"hdfs:///user/model/RandomForestRegression"}' '{"OutputHiveTable":"RandomForestRegressionPre"}'

聚类

  1. BisectingKMeans
参数名称 参数类型 参数描述 默认值 是否必选
K int The number of clusters to infer. Must be > 1. 4 True
Seed long Random seed. (default: 566573821) 566573821 False
PredictionCol String Param for prediction column name. prediction True
FeaturesCol String Column of features. features True
maxInter int Param for maximum number of iterations (>= 0). (default: 20) 20 False
minDivisibleClusterSize int Minnum size of divisibleCluster (default: 1) 1 False

1.1 测试用例
clustering.BisectingKMeans
1.2 提交参数
spark-submit --class grimoire.ml.clustering.conjure.BisectingKMeans --master local[4] 'grimoire-assembly-0.1.0.jar' '{"InputHiveTable":"BisectingKMeans"}' '{"LabelCol":"label","FeaturesCol":"features","PredictionCol":"prediction","K":2}' '{"OutputModelTarget":"hdfs:///user/model/BisectingKMeans","Overwrite":true}'

2.1 测试用例
clustering.BisectingKMeansPre
2.2 提交参数
spark-submit --class grimoire.ml.clustering.conjure.BisectingKMeansPre --master local[4] 'grimoire-assembly-0.1.0.jar' '{"InputHiveTable":"BisectingKMeans"}' '{"InputModelSource":"hdfs:///user/model/BisectingKMeans"}' '{"OutputHiveTable":"BisectingKMeansPre"}'

协同过滤

  1. ALS
参数名称 参数类型 参数描述 默认值 是否必选
MaxIter int Max iteration of train (default: 10) 10 False
RegParam double Regularize parameter (default: 0.1) 0.1 False
UserCol String Param for the column name for user ids. user True
ItemCol String Param for the column name for item ids. item True
RatingCol String Param for the column name for ratings. rating True
PredictionCol String Param for prediction column name. prediction True
Alpha double Param for the alpha parameter in the implicit preference formulation (nonnegative). (default: 1.0) 1.0 False
CheckpointInterval int Param for set checkpoint interval (>= 1) or disable checkpoint (-1). (default: 10) 10 False
FinalStorageLevel String (default: MEMORY_AND_DISK) MEMORY_AND_DISK False
Nonnegative boolean Param for whether to apply nonnegativity constraints. (default: false) false False
NumUserBlocks int Param for number of blocks. (default: 10) 10 False
NumItemBlocks int Param for number of item blocks (positive). (default: 10) 10 False
Rank int Param for rank of the matrix factorization (positive). (default: 10) 10 False
Seed long Random seed. (default: 1994790107) 1994790107 False
ImplicitPrefs boolean Param to decide whether to use implicit preference. (default: false) false False

1.1 测试用例
filtering.ALS
1.2 提交参数
spark-submit --class grimoire.ml.filtering.conjure.ALS --master local[4] 'grimoire-assembly-0.1.0.jar' '{"InputHiveTable":"ALS"}' '{"MaxIter":5,"RegParam":0.01,"UserCol":"user","ItemCol":"product","RatingCol":"rating","PredictionCol":"prediction"}' '{"OutputModelTarget":"hdfs:///user/model/ALS","Overwrite":true}'

2.1 测试用例
filtering.ALSPre
2.2 提交参数
spark-submit --class grimoire.ml.filtering.conjure.ALSPre --master local[4] 'grimoire-assembly-0.1.0.jar' '{"InputHiveTable":"ALS"}' '{"InputModelSource":"hdfs:///user/model/ALS"}' '{"OutputHiveTable":"ALSPre"}'

统计

  1. Correlations
参数名称 参数类型 参数描述 默认值 是否必选
InputCols Seq names of input columns (default: null) None False
CorrelationMethod String Correlation method(default: pearson; alternative: spearman). (default: pearson) pearson False

1.1 测试用例
statistics.ALS
1.2 提交参数
spark-submit --class grimoire.ml.statistics.conjure.Correlations --master local[4] 'grimoire-assembly-0.1.0.jar' '{"InputHiveTable":"iris"}' '{"InputCols":["f1","f2","f3","f4"],"CorrelationMethod":"pearson"}' '{"OutPutJsonTarget":"hdfs:///user/example/jsonss.txt"}'

  1. StatisticalSummary
参数名称 参数类型 参数描述 默认值 是否必选
InputCols Seq Param for input column names. None True
rowLabels Seq the labels of rows. (default: List(count, max, min, mean, normL1, normL2, numNonzeros, variance)) List(count, max, min, mean, normL1, normL2, numNonzeros, variance) False
ColLabels Seq the labels of columns. None True
transposed boolean whether or not is transposed (default: true) true False
numCols int the number of rows (default: null) None False
numRows int the number of columns (default: null) None False

2.1 测试用例
statistics.StatisticalSummary
2.2 提交参数
spark-submit --class grimoire.ml.statistics.conjure.StatisticalSummary --master local[4] 'grimoire-assembly-0.1.0.jar' '{"InputHiveTable":"iris"}' '{"InputCols":["f1","f2","f3","f4"],"ColLabels":["f1","f2","f3","f4"]}' '{"OutPutJsonTarget":"hdfs:///user/example/jsonss"}'

模型评估

  1. BinaryClassificationMetrics
参数名称 参数类型 参数描述 默认值 是否必选
LabelCol String Column name of label. label True
PredictionCol String Param for prediction column name. prediction True

1.1 测试用例
evaluate.BinaryClassificationMetrics
1.2 提交参数
spark-submit --class grimoire.ml.evaluate.conjure.BinaryClassificationMetrics --master local[4] 'grimoire-assembly-0.1.0.jar' 'RandomForest' '{"InputHiveTable":"DecisionTree"}' '{"InputModelSource":"hdfs:///user/model/RandomForest"}' '{"LabelCol":"label","PredictionCol":"prediction"}' '{"OutPutJsonTarget":"hdfs:///user/example/jsonss"}'

  1. RegressionMetrics
参数名称 参数类型 参数描述 默认值 是否必选
LabelCol String Column name of label. label True
PredictionCol String Param for prediction column name. prediction True

1.1 测试用例
evaluate.RegressionMetrics
1.2 提交参数
spark-submit --class grimoire.ml.evaluate.conjure.RegressionMetrics --master local[4] 'grimoire-assembly-0.1.0.jar' 'LinearRegression' '{"InputHiveTable":"REGRESSION"}' '{"InputModelSource":"hdfs:///user/model/REGRESSIONMOD"}' '{"LabelCol":"label","PredictionCol":"prediction"}' '{"OutPutJsonTarget":"hdfs:///user/example/jsonss"}'

你可能感兴趣的:(sparkMlib_doc_1.0)

  • sparkMlib_doc_1.0 我是一只会飞的羊
    模型输入输出对应关系输入表（hive）——模型参数——输出模型（hdfs）DecisionTreeGBTCLogisticRegressionNaiveBayesRandomForestBisectingKMeansIDFTrainALSDecisionTreeRegressionLinearRegressionRandomForestRegression示例：'{"InputHiveTable"
  • 二分查找排序算法 周凡杨 java二分查找排序算法折半
    一：概念 二分查找又称 折半查找（ 折半搜索/ 二分搜索），优点是比较次数少，查找速度快，平均性能好；其缺点是要求待查表为有序表，且插入删除困难。因此，折半查找方法适用于不经常变动而 查找频繁的有序列表。首先，假设表中元素是按升序排列，将表中间位置记录的关键字与查找关键字比较，如果两者相等，则查找成功；否则利用中间位置记录将表 分成前、后两个子表，如果中间位置记录的关键字大于查找关键字，则进一步
  • java中的BigDecimal bijian1013 javaBigDecimal
            在项目开发过程中出现精度丢失问题，查资料用BigDecimal解决，并发现如下这篇BigDecimal的解决问题的思路和方法很值得学习，特转载。         原文地址：http://blog.csdn.net/ugg/article/de
  • Shell echo命令详解 daizj echoshell
    Shell echo命令 Shell 的 echo 指令与 PHP 的 echo 指令类似，都是用于字符串的输出。命令格式： echo string 您可以使用echo实现更复杂的输出格式控制。 1.显示普通字符串: echo "It is a test" 这里的双引号完全可以省略，以下命令与上面实例效果一致： echo Itis a test 2.显示转义
  • Oracle DBA 简单操作 周凡杨 oracle dba sql
    --执行次数多的SQL select sql_text,executions from (      select sql_text,executions from v$sqlarea order by executions desc      ) where rownum<81;  &nb
  • 画图重绘 朱辉辉33 游戏
      我第一次接触重绘是编写五子棋小游戏的时候，因为游戏里的棋盘是用线绘制的，而这些东西并不在系统自带的重绘里，所以在移动窗体时，棋盘并不会重绘出来。所以我们要重写系统的重绘方法。   在重写系统重绘方法时，我们要注意一定要调用父类的重绘方法，即加上super.paint(g)，因为如果不调用父类的重绘方式，重写后会把父类的重绘覆盖掉，而父类的重绘方法是绘制画布，这样就导致我们
  • 线程之初体验 西蜀石兰 线程
    一直觉得多线程是学Java的一个分水岭，懂多线程才算入门。 之前看《编程思想》的多线程章节，看的云里雾里，知道线程类有哪几个方法，却依旧不知道线程到底是什么？书上都写线程是进程的模块，共享线程的资源，可是这跟多线程编程有毛线的关系，呜呜。。。 线程其实也是用户自定义的任务，不要过多的强调线程的属性，而忽略了线程最基本的属性。 你可以在线程类的run()方法中定义自己的任务，就跟正常的Ja
  • linux集群互相免登陆配置 林鹤霄 linux
    配置ssh免登陆 1、生成秘钥和公钥    ssh-keygen -t rsa 2、提示让你输入，什么都不输，三次回车之后会在~下面的.ssh文件夹中多出两个文件id_rsa 和 id_rsa.pub    其中id_rsa为秘钥，id_rsa.pub为公钥，使用公钥加密的数据只有私钥才能对这些数据解密    c
  • mysql : Lock wait timeout exceeded; try restarting transaction aigo mysql
    原文：http://www.cnblogs.com/freeliver54/archive/2010/09/30/1839042.html   原因是你使用的InnoDB   表类型的时候, 默认参数:innodb_lock_wait_timeout设置锁等待的时间是50s, 因为有的锁等待超过了这个时间,所以抱错.   你可以把这个时间加长,或者优化存储
  • Socket编程 基本的聊天实现。 alleni123 socket
    public class Server { //用来存储所有连接上来的客户 private List<ServerThread> clients; public static void main(String[] args) { Server s = new Server(); s.startServer(9988); } publi
  • 多线程监听器事件模式(一个简单的例子) 百合不是茶 线程监听模式
        多线程的事件监听器模式   监听器时间模式经常与多线程使用,在多线程中如何知道我的线程正在执行那什么内容,可以通过时间监听器模式得到        创建多线程的事件监听器模式 思路:    1, 创建线程并启动,在创建线程的位置设置一个标记     2,创建队
  • spring InitializingBean接口 bijian1013 javaspring
    spring的事务的TransactionTemplate，其源码如下： public class TransactionTemplate extends DefaultTransactionDefinition implements TransactionOperations, InitializingBean{ ... } TransactionTemplate继承了DefaultT
  • Oracle中询表的权限被授予给了哪些用户 bijian1013 oracle数据库权限
            Oracle查询表将权限赋给了哪些用户的SQL，以备查用。 select t.table_name as "表名", t.grantee as "被授权的属组", t.owner as "对象所在的属组"
  • 【Struts2五】Struts2 参数传值 bit1129 struts2
    Struts2中参数传值的3种情况 1.请求参数绑定到Action的实例字段上 2.Action将值传递到转发的视图上 3.Action将值传递到重定向的视图上   一、请求参数绑定到Action的实例字段上以及Action将值传递到转发的视图上 Struts可以自动将请求URL中的请求参数或者表单提交的参数绑定到Action定义的实例字段上，绑定的规则使用ognl表达式语言
  • 【Kafka十四】关于auto.offset.reset[Q/A] bit1129 kafka
    I got serveral questions about  auto.offset.reset. This configuration parameter governs how  consumer read the message from  Kafka when  there is no initial offset in ZooKeeper or
  • nginx gzip压缩配置 ronin47 nginx gzip 压缩范例
    nginx gzip压缩配置 更多 0 nginx gzip 配置   随着nginx的发展，越来越多的网站使用nginx，因此nginx的优化变得越来越重要，今天我们来看看nginx的gzip压缩到底是怎么压缩的呢？ gzip(GNU-ZIP)是一种压缩技术。经过gzip压缩后页面大小可以变为原来的30%甚至更小，这样，用
  • java-13.输入一个单向链表，输出该链表中倒数第 k 个节点 bylijinnan java
    two cursors. Make the first cursor go K steps first. /* * 第 13 题：题目：输入一个单向链表，输出该链表中倒数第 k 个节点 */ public void displayKthItemsBackWard(ListNode head,int k){ ListNode p1=head,p2=head;
  • Spring源码学习-JdbcTemplate queryForObject bylijinnan javaspring
    JdbcTemplate中有两个可能会混淆的queryForObject方法： 1. Object queryForObject(String sql, Object[] args, Class requiredType) 2. Object queryForObject(String sql, Object[] args, RowMapper rowMapper) 第1个方法是只查
  • [冰川时代]在冰川时代,我们需要什么样的技术? comsci 技术
         看美国那边的气候情况....我有个感觉...是不是要进入小冰期了?      那么在小冰期里面...我们的户外活动肯定会出现很多问题...在室内呆着的情况会非常多...怎么在室内呆着而不发闷...怎么用最低的电力保证室内的温度.....这都需要技术手段...   &nb
  • js 获取浏览器型号 cuityang js浏览器
    根据浏览器获取iphone和apk的下载地址 <!DOCTYPE html> <html> <head>     <meta charset="utf-8" content="text/html"/>     <meta name=
  • C# socks5详解 转 dalan_123 socketC#
    http://www.cnblogs.com/zhujiechang/archive/2008/10/21/1316308.html  这里主要讲的是用.NET实现基于Socket5下面的代理协议进行客户端的通讯，Socket4的实现是类似的，注意的事，这里不是讲用C#实现一个代理服务器，因为实现一个代理服务器需要实现很多协议，头大，而且现在市面上有很多现成的代理服务器用，性能又好，
  • 运维 Centos问题汇总 dcj3sjt126com 云主机
    一、sh 脚本不执行的原因 sh脚本不执行的原因 只有2个 1.权限不够 2.sh脚本里路径没写完整。   二、解决You have new mail in /var/spool/mail/root 修改/usr/share/logwatch/default.conf/logwatch.conf配置文件 MailTo = MailFrom   三、查询连接数
  • Yii防注入攻击笔记 dcj3sjt126com sqlWEB安全yii
    网站表单有注入漏洞须对所有用户输入的内容进行个过滤和检查，可以使用正则表达式或者直接输入字符判断，大部分是只允许输入字母和数字的，其它字符度不允许；对于内容复杂表单的内容，应该对html和script的符号进行转义替换：尤其是<,>,',"",&这几个符号 这里有个转义对照表： http://blog.csdn.net/xinzhu1990/articl
  • MongoDB简介[一] eksliang mongodbMongoDB简介
    MongoDB简介 转载请出自出处：http://eksliang.iteye.com/blog/2173288 1.1易于使用        MongoDB是一个面向文档的数据库，而不是关系型数据库。与关系型数据库相比，面向文档的数据库不再有行的概念，取而代之的是更为灵活的“文档”模型。        另外，不
  • zookeeper windows 入门安装和测试 greemranqq zookeeper安装分布式
    一、序言       以下是我对zookeeper 的一些理解：      zookeeper 作为一个服务注册信息存储的管理工具，好吧，这样说得很抽象，我们举个“栗子”。       栗子1号：       假设我是一家KTV的老板，我同时拥有5家KTV，我肯定得时刻监视
  • Spring之使用事务缘由(2-注解实现) ihuning spring
      Spring事务注解实现   1. 依赖包：     1.1 spring包：           spring-beans-4.0.0.RELEASE.jar           spring-context-4.0.0.
  • iOS App Launch Option 啸笑天 option
    iOS 程序启动时总会调用application:didFinishLaunchingWithOptions:，其中第二个参数launchOptions为NSDictionary类型的对象，里面存储有此程序启动的原因。   launchOptions中的可能键值见UIApplication Class Reference的Launch Options Keys节 。 1、若用户直接
  • jdk与jre的区别（_） macroli javajvmjdk
    简单的说JDK是面向开发人员使用的SDK，它提供了Java的开发环境和运行环境。SDK是Software Development Kit 一般指软件开发包，可以包括函数库、编译程序等。  JDK就是Java Development Kit JRE是Java Runtime Enviroment是指Java的运行环境，是面向Java程序的使用者，而不是开发者。 如果安装了JDK，会发同你
  • Updates were rejected because the tip of your current branch is behind qiaolevip 学习永无止境每天进步一点点众观千象git
    $ git push joe prod-2295-1 To [email protected]:joe.le/dr-frontend.git ! [rejected] prod-2295-1 -> prod-2295-1 (non-fast-forward) error: failed to push some refs to '[email protected]
  • [一起学Hive]之十四-Hive的元数据表结构详解 superlxw1234 hivehive元数据结构
    关键字：Hive元数据、Hive元数据表结构   之前在 “[一起学Hive]之一–Hive概述，Hive是什么”中介绍过，Hive自己维护了一套元数据，用户通过HQL查询时候，Hive首先需要结合元数据，将HQL翻译成MapReduce去执行。 本文介绍一下Hive元数据中重要的一些表结构及用途，以Hive0.13为例。   文章最后面，会以一个示例来全面了解一下，
  • Spring 3.2.14，4.1.7，4.2.RC2发布 wiselyman Spring 3
      Spring 3.2.14、4.1.7及4.2.RC2于6月30日发布。   其中Spring 3.2.1是一个维护版本(维护周期到2016-12-31截止)，后续会继续根据需求和bug发布维护版本。此时，Spring官方强烈建议升级Spring框架至4.1.7 或者将要发布的4.2 。   其中Spring 4.1.7主要包含这些更新内容。
按字母分类： ABCDEFGHIJKLMNOPQRSTUVWXYZ其他