Hi,
I’m afraid I don’t know that part well enough. What’s the percentage in
slowdown? (7 seconds alone doesn’t say anything)
Maybe Till (in cc) knows more since he used to work on the ML part.
Best,
Aljoscha
> On 6. Jun 2017, at 17:45, Borja <borja.r.mad...@gmail.com> wrote:
>
> *Thank so much Aljoscha* :)
> I was stucked in this point. I didn't know that the print or collect method
> collecting all the data in one place.
>
> The execution time has dropped a lot.
> However, I still get that Flink is slower (just for 7 seconds).
>
> I really think I'm not getting all the performance out of Flink.
> Because Flink draws the execution in a cyclic dependency graph meanwhile
> Spark uses a DAG,
> so it's clear that the Flin's way results in superior scalability and
> performance compared to DAG approach.
>
> So... Which is the problem with my code?
>
> //Read data
> val data: DataSet[org.apache.flink.ml.common.LabeledVector] =
> MLUtils.readLibSVM(benv, "/inputPath/_.libsvm")
>
> // Create multiple linear regression learner
> val mlr = MultipleLinearRegression()
>
> val model = mlr.fit(data)
>
> data.writeAsText("file:///outputPath")
>
> benv.execute()
>
>
>
> --
> View this message in context:
> http://apache-flink-user-mailing-list-archive.2336050.n4.nabble.com/Methods-that-trigger-execution-tp12972p13537.html
> Sent from the Apache Flink User Mailing List archive. mailing list archive at
> Nabble.com.
