Compute rolling mean with ojalgo

Question

What is best way to compute rolling mean using ojalgo?

First SampleSet is adding every number as many times as windows size is.

Second the code intent is not evident.

I have:

    final Array1D<Double> doubles = Array1D.factory(Primitive32Array.FACTORY).makeFilled(10, new Uniform());
    System.out.println(doubles.toString());

    final Array1D<Double> rollingMedian = Array1D.PRIMITIVE32.makeZero(doubles.size());

    rollingMedian.set(0, doubles.get(0));
    System.out.printf("%1$s -> %1$s\n", doubles.get(0));

    final Array1D<Double> someSamples2 = doubles.subList(0, 2);
    final double mean2 = SampleSet.wrap(someSamples2).getMean();
    rollingMedian.set(1, mean2);
    System.out.printf("%s -> %s\n", someSamples2.toString(), mean2);

    for (int i = 2; i < doubles.length; i++) {
        final Array1D<Double> someSamples = doubles.subList(i - 2, i + 1);
        final double mean = SampleSet.wrap(someSamples).getMean();
        rollingMedian.set(i, mean);
        System.out.printf("%s -> %s\n", someSamples.toString(), mean2);
    }

    System.out.println(rollingMedian.toString());

Which is not nicest...

apete · Accepted Answer · 2018-04-26T15:40:09.450

1

Not sure which would be the nicest way, but here's one alternatives:

    for (int i = 0; i < doubles.length; i++) {
        int first = Math.max(0, i - 2);
        int limit = i + 1;
        double mean = doubles.aggregateRange(first, limit, Aggregator.AVERAGE);
        rollingMedian.set(i, mean);
    }

edited Apr 26 '18 at 15:40

answered Apr 26 '18 at 15:21

apete

1,250
1
10
16

I do not see Aggregator.AVERAGE and Array1D does not implement Access1D.Aggregatable – user482745 Apr 26 '18 at 15:52
1

If you don't see Aggregator.AVERAGE you're not using the latest version - it was just added with v45.1. – apete Apr 26 '18 at 17:17
How would you like to do this? – apete Apr 26 '18 at 17:18
I may create pull request for adding rollaply function to Array1D or appropriate interface. – user482745 Apr 28 '18 at 05:47

Compute rolling mean with ojalgo

1 Answers1