开发者

Using LINQ to create an IEnumerable<> of delta values

I've got a list of timestamps (in ticks), and from this list I'd like to create another one that represents开发者_StackOverflow社区 the delta time between entries.

Let's just say, for example, that my master timetable looks like this:

  1. 10
  2. 20
  3. 30
  4. 50
  5. 60
  6. 70

What I want back is this:

  1. 10
  2. 10
  3. 20
  4. 10
  5. 10

What I'm trying to accomplish here is detect that #3 in the output table is an outlier by calculating the standard deviation. I've not taken statistics before, but I think if I look for the prevalent value in the output list and throw out anything outside of 1 sigma that this will work adequately for me.

I'd love to be able to create the output list with a single LINQ query, but I haven't figured it out yet. Currently I'm just brute forcing it with a loop.


If you are running .NET 4.0, this should work fine:

var deltas = list.Zip(list.Skip(1), (current, next) => next - current);

Apart from the multiple enumerators, this is quite efficient; it should work well on any kind of sequence.

Here's an alternative for .NET 3.5:

var deltas = list.Skip(1)
                 .Select((next, index) => next - list[index]);

Obviously, this idea will only be efficient when the list's indexer is employed. Modifying it to use ElementAt may not be a good idea: quadratic run-time will occur for non IList<T> sequences. In this case, writing a custom iterator is a good solution.

EDIT: If you don't like the Zip + Skip(1) idea, writing an extension such as this (untested) maybe useful in these sorts of circumstances:

public class CurrentNext<T>
{
    public T Current { get; private set; }
    public T Next { get; private set; }

    public CurrentNext(T current, T next)
    {
        Current = current;
        Next = next;
    }
}

...

public static IEnumerable<CurrentNext<T>> ToCurrentNextEnumerable<T>(this IEnumerable<T> source)
{
    if (source == null)
        throw new ArgumentException("source");

    using (var source = enumerable.GetEnumerator())
    {
        if (!enumerator.MoveNext())
            yield break;

        T current = enumerator.Current;

        while (enumerator.MoveNext())
        {
            yield return new CurrentNext<T>(current, enumerator.Current);
            current = enumerator.Current;
        }
    }
}

Which you could then use as:

var deltas = list.ToCurrentNextEnumerable()
                 .Select(c=> c.Next - c.Current);


You can use Ani's answer:-

var deltas = list.Zip(list.Skip(1), (current, next) => next - current);

With a super-simple implementation of the Zip extension method:-

public static IEnumerable<TResult> Zip<TFirst, TSecond, TResult>(
  this IEnumerable<TFirst> first,
  IEnumerable<TSecond> second,
  Func<TFirst, TSecond, TResult> func)
{
  var ie1 = first.GetEnumerator();
  var ie2 = second.GetEnumerator();

  while (ie1.MoveNext() && ie2.MoveNext())
    yield return func(ie1.Current, ie2.Current);
}

That'll work with 3.5.


This should do the trick:

static IEnumerable<int> GetDeltas(IEnumerable<int> collection)
{
    int? previous = null;

    foreach (int value in collection)
    { 
        if (previous != null)
        {
            yield return value - (int)previous;
        }
        previous = value;
    }
}

Now you can call your collection like this:

var masterTimetable = GetMasterTimeTable();

var deltas = GetDeltas(masterTimetable);

It's not really LINQ, but will effectively do the trick.


It looks like there are sufficient answers to get you going already, but I asked a similar question back in the spring:

How to zip one ienumerable with itself

In the responses to my question, I learned about "Pairwise" and "Pairwise"

As I recall, explicitly implementing your own "Pairwise" enumerator does mean that you iterate through you list exactly once whereas implementing "Pairwise" in terms of .Zip + .Skip(1) means that you will ultimately iterate over your list twice.

In my post I also include several examples of geometry (operating on lists of points) processing code such as Length/Distance, Area, Centroid.


Not that I recommend this, but totally abusing LINQ the following would work:

var vals = new[] {10, 20, 30, 50, 60, 70};

int previous = 0;
var newvals = vals.Select(i =>
                            {
                                int dif = i - previous;
                                previous = i;
                                return dif;
                            });
foreach (var newval in newvals)
{
    Console.WriteLine(newval);
}


One liner for you:

int[] i = new int[] { 10, 20, 30, 50, 60, 70 };
IEnumerable<int> x = Enumerable.Range(1, i.Count()-1).Select(W => i[W] - i[W - 1]);


LINQ is not really designed for what you're trying to do here, because it usually evaluates value by value, much like an extremely efficient combination of for-loops. You'd have to know your current index, something you don't, without some kind of workaround.

0

上一篇:

下一篇:

精彩评论

暂无评论...
验证码 换一张
取 消

最新问答

问答排行榜