4

I use Twitter anomaly detection algorithm in my project. For this purpose I use Rserve library to run R code in my Java application.

My Java code:

RConnection connection = new RConnection();
connection.voidEval("library(AnomalyDetection)");
connection.eval("res <- AnomalyDetectionTs(data.frame(/*list of timestamps*/,/*list of values*/), direction='both', plot=FALSE, longterm=TRUE)");

And, as a result, I got this output:

    $anoms
              timestamp    anoms
1   1980-09-25 16:05:00  21.3510
2   1980-09-29 06:40:00 193.1036
3   1980-09-29 21:44:00 148.1740

To get results now I'm using this not nice solution: connection.eval("write.csv(res[['anoms']],file='anom.csv')");

Then I open this file in Java and parse the results.

So, how to get the output results in Java using Rserve possibilities for data.frame structure?

Improve this question

1 Answer 1

Reset to default
4

Simply write the R command such that it returns the desired result back to Java:

RList l = c.eval("AnomalyDetectionTs(data, direction='both',
                 plot=FALSE, longterm=TRUE)$anoms").asList();

What you get is the data frame (as a list) with the two variables timestamp and anoms.

However, AnomalyDetectionTs returns dates in a horribly annoying and inefficient format so you may want to actually return a more sane result which is easier to work with in Java, e.g.:

RList l = c.eval("{ res <- AnomalyDetectionTs(data, direction='both', plot=FALSE,
                          longterm=TRUE)$anoms;
                  list(as.POSIXct(res$timestamp), res$anoms) }").asList();
double ts[] = l.at(0).asDoubles();
double anom[] = l.at(1).asDoubles();
for (int i = 0; i < ts.length; i++)
    System.out.println(new java.util.Date((long)(ts[i]*1000.0)) + ": " + anom[i]);

PS: the right place for Rserve questions is the stats-rosuda-devel mailing list which will give you answers much faster.

Improve this answer
Sign up to request clarification or add additional context in comments.

2 Comments

Excellent. Thank you
This works like a charm, but is there a reason we are not using RVector in place of RList?

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Start asking to get answers

Find the answer to your question by asking.

Ask question

Explore related questions

See similar questions with these tags.