在mllib LinearRegressionWithSGD中,我无法控制numIterations,为什么?

时间:2016-11-10 09:18:45

标签: scala apache-spark machine-learning apache-spark-mllib

我有1毫升任务,34461个样本,我使用LinearRegressionWithSGD来训练模型。但是两个训练结果(MSE和相关)的结果是相同的:

列车模型(setNumIterations( 5 ))

val linearRegressionWithSGD: LinearRegressionWithSGD = new LinearRegressionWithSGD()
linearRegressionWithSGD.setIntercept(true)
linearRegressionWithSGD.optimizer.setStepSize(0.1)
linearRegressionWithSGD.optimizer.setNumIterations(5)

列车模型(setNumIterations( 10000 ))

val linearRegressionWithSGD: LinearRegressionWithSGD = new LinearRegressionWithSGD()
linearRegressionWithSGD.setIntercept(true)
linearRegressionWithSGD.optimizer.setStepSize(0.1)
linearRegressionWithSGD.optimizer.setNumIterations(10000)

我无法控制迭代次数,但为什么?

0 个答案:

没有答案