您好,登錄后才能下訂單哦!
其實predictionio 也是支持隨機森林算法的,可惜官方的例子沒有出現,我簡單寫了一點,您也可以參考這個地址:
https://github.com/runapal5/PredictionIO-DefectPrediction/blob/master/engine.json
public class RandomForestAlgorithm extends P2LJavaAlgorithm<PreparedData, RandomForestModel, Query, PredictedResult> {
/**
*
*/
private static final long serialVersionUID = 1L;
private final RandomForestAlgorithmParams ap;
public RandomForestAlgorithm(RandomForestAlgorithmParams ap) {
this.ap = ap;
}
@Override
public RandomForestModel train(SparkContext sc, PreparedData preparedData) {
String datapath = "/root/whk/company.txt";
JavaRDD<LabeledPoint> treedata = MLUtils.loadLibSVMFile(sc, datapath).toJavaRDD();
Map<Integer,Integer> categoricalFeaturesInfo = new HashMap<Integer,Integer>();
return RandomForest.trainClassifier(treedata, ap.getNumClasses(), categoricalFeaturesInfo, ap.getNumTrees(),
ap.getFeatureSubsetStrategy(),ap.getImpurity(),ap.getMaxDepth(),ap.getMaxBins(),12345);
}
@Override
public PredictedResult predict(RandomForestModel model, final Query query) {
double[] b={12,0,0};
model.predict(Vectors.dense(b));
return null;
}
}
免責聲明:本站發布的內容(圖片、視頻和文字)以原創、轉載和分享為主,文章觀點不代表本網站立場,如果涉及侵權請聯系站長郵箱:is@yisu.com進行舉報,并提供相關證據,一經查實,將立刻刪除涉嫌侵權內容。