java 朴素贝叶斯_朴素贝叶斯算法java实现（多项式模型）

/*** 对分类结果进行比较，得出概率最大的类

*@paramclassifyResult

*@return

*/

public static String getClassifyResultName(MapclassifyResult){

String classifyName= "";if(classifyResult.isEmpty()){returnclassifyName;

}

BigDecimal result= new BigDecimal(0);

Set classifyResultSet =classifyResult.keySet();for(String classifyResultSetString : classifyResultSet){if (classifyResult.get(classifyResultSetString).compareTo(result) >= 1){

result=classifyResult.get(classifyResultSetString);

classifyName=classifyResultSetString;

}

}returnclassifyName;

}/*** 统计给定类别下的单词总数(带词频计算)

*@paramcategoryLabel 指定类别参数

*@return

*/

public staticLong categoryWordCount(String categoryLabel){

Long sum= 0L;

Map> categoryWordMap =allTrainFileSegsMap.get(categoryLabel);if (categoryWordMap == null){returnsum;

}

Set categoryWordMapKeySet =categoryWordMap.keySet();for(String categoryLabelString : categoryWordMapKeySet){

Map categoryWordMapDataMap =categoryWordMap.get(categoryLabelString);

List> dataWordMapList = new ArrayList>(categoryWordMapDataMap.entrySet());for (int i=0; i

sum+=dataWordMapList.get(i).getValue();

}

}returnsum;

}/*** 获取训练样本所有词的总数(词总数计算是带上词频的，也就是可以重复算数)

*@return

*/

public staticLong getAllTrainCategoryWordsCount(){

Long sum= 0L;//获取所有分类

Set categoryLabels =allTrainFileSegsMap.keySet();//循环相加每个类下的词总数

for(String categoryLabel : categoryLabels){

sum+=categoryWordCount(categoryLabel);

}returnsum;

}/*** 获取训练样本下各个类别不重复词的总词数，区别于getAllTrainCategoryWordsCount()方法，此处计算不计算词频

* 备注：此处并不是严格意义上的进行全量词表生成后的计算，也就是加入类别1有"中国=6"、类别2有"中国=2"，总词数算中国两次，

* 也就是说，我们在计算的时候并没有生成全局词表(将所有词都作为出现一次)

*@return

*/

public staticLong getAllTrainCategoryWordCount(){

Long sum= 0L;//获取所有分类

Set categoryLabels =allTrainFileSegsMap.keySet();for(String cateGoryLabelsLabel : categoryLabels){

Map> categoryWordMap =allTrainFileSegsMap.get(cateGoryLabelsLabel);

List>> categoryWordMapList = new ArrayList>>(categoryWordMap.entrySet());for (int i=0; i

sum+=categoryWordMapList.get(i).getValue().size();

}

}returnsum;

}/*** 计算测试数据的每个单词在每个类下出现的总数

*@paramword

*@paramcategoryLabel

*@return

*/

public staticLong wordInCategoryCount(String word, String categoryLabel){

Long sum= 0L;

Map> categoryWordMap =allTrainFileSegsMap.get(categoryLabel);

Set categoryWordMapKeySet =categoryWordMap.keySet();for(String categoryWordMapKeySetFile : categoryWordMapKeySet){

Map categoryWordMapDataMap =categoryWordMap.get(categoryWordMapKeySetFile);

Integer value=categoryWordMapDataMap.get(word);if (value!=null && value>0){

sum+=value;

}

}returnsum;

}/*** 获取所有分类类别

*@return

*/

public SetgetAllCategory(){returnallTrainFileSegsMap.keySet();

}