Help comparing float member variables using Comparators

Sheehan Alam picture Sheehan Alam · Sep 14, 2010 · Viewed 19.2k times · Source

I am able to compare Strings fine, but would like to know how I can rank floating point numbers?

getChange() returns a String. I want to be able to sort descending. How can I do this?

UPDATE:

package org.stocktwits.helper;

import java.util.Comparator;

import org.stocktwits.model.Quote;

public class ChangeComparator implements Comparator<Quote>
{
    public int compare(Quote o1, Quote o2) {
        float change1 = Float.valueOf(o1.getChange());
        float change2 = Float.valueOf(o2.getChange());

        if (change1 < change2) return -1;
        if (change1 == change2) return 0; // Fails on NaN however, not sure what you want
        if (change2 > change2) return 1;
    }
}

I am getting the compile time error:

This method must return a result of type int    ChangeComparator.java   

Answer

clstrfsck picture clstrfsck · Sep 14, 2010

How about this:

public class ChangeComparator implements Comparator<Quote>
{
    public int compare(Quote o1, Quote o2) {
        Float change1 = Float.valueOf(o1.getChange());
        Float change2 = Float.valueOf(o2.getChange());
        return change1.compareTo(change2);
    }
}

Note that Java 1.4 introduced Float#compare(float, float) (and an equivalent in Double), which can be pretty much used directly:

public class ChangeComparator implements Comparator<Quote>
{
    public int compare(Quote o1, Quote o2) {
        return Float.compare(o1.getChange(), o2.getChange());
    }
}

(After editing, I notice that @BorislavGizdov has mentioned this in his answer already.)


Also worth noting that Java 8 Comparator#comparing(...) and Comparator#comparingDouble(...) provide a straightforward way of constructing these comparators directly.

Comparator<Quote> changeComparator = Comparator.comparing(Quote::getChange);

Will compare using boxed Float values.

Comparator<Quote> changeComparator = Comparator.comparingDouble(Quote::getChange);

Will compare using float values promoted to double values.

Given that there is no Comparator#comparingFloat(...), my preference would be to use the comparingDouble(...) method, as this only involves primitive type conversion, rather than boxing.