如何计算C＃中某些时间跨度的平均值

Question

I already search this question but unfortunately couldn't find proper answer.

I want to calculate the average of time spent on doing something in 10 different day and I have 10 datetimepicker for start time and also 10 datetimepicker for end time for each day (in total 20 datetimepicker ).

now I want to get the average of time spent for work in these 10 days.

this is what I've done for calculating timespan in each day and now I don't know how to calculate average if these timespans

of course I want to know is there any shorter way to get the job done?

DateTime Dt1 = dateTimePicker1.Value;
DateTime Dt2 = dateTimePicker2.Value;
.
.
.
DateTime Dt20 = dateTimePicker20.Value;

TimeSpan Day1 = DateTime.Parse(Dt11.TimeOfDay.ToString()).Subtract(DateTime.Parse(Dt1.TimeOfDay.ToString()));
TimeSpan Day2 = DateTime.Parse(Dt12.TimeOfDay.ToString()).Subtract(DateTime.Parse(Dt2.TimeOfDay.ToString()));
.
.
.

TimeSpan Day10 = DateTime.Parse(Dt20.TimeOfDay.ToString()).Subtract(DateTime.Parse(Dt10.TimeOfDay.ToString()));

I want to find average of Day1 to Day10

Answer 1

Given an IEnumerable<TimeSpan> you can average them with this extension method:

public static TimeSpan Average(this IEnumerable<TimeSpan> spans) => TimeSpan.FromSeconds(spans.Select(s => s.TotalSeconds).Average());

So, convert your results to a List :

var durs = new List<TimeSpan>();
durs.Add(Dt11.TimeOfDay.Subtract(Dt1.TimeOfDay));
durs.Add(Dt12.TimeOfDay.Subtract(Dt2.TimeOfDay));
.
.
.

durs.Add(Dt20.TimeOfDay.Subtract(Dt10.TimeOfDay));

Now compute the average:

var avgDurs = durs.Average();

PS: Created an Aggregate version of @MattJohnson's answer:

public static TimeSpan Mean(this IEnumerable<TimeSpan> source) => TimeSpan.FromTicks(source.Aggregate((m: 0L, r: 0L, n: source.Count()), (tm, s) => {
        var r = tm.r + s.Ticks % tm.n;
        return (tm.m + s.Ticks / tm.n + r / tm.n, r % tm.n, tm.n);
    }).m);

Answer 2

You can find average by instantiating a new TimeSpan based on number of ticks.

    var time_spans = new List<TimeSpan>() { new TimeSpan(24, 10, 0), new TimeSpan(12, 0, 45), new TimeSpan(23, 30, 0), new TimeSpan(11, 34, 0) };

    var average = new TimeSpan(Convert.ToInt64(time_spans.Average(t => t.Ticks)));

     Console.WriteLine(average); 

result 17:48:41.2500000

Answer 3

Just to follow up on NetMage's perfectly good answer, note that his Average extension method is using .TotalSeconds , which returns a double of whole and fractional seconds, with millisecond precision. That is probably fine if you are taking values from time-pickers, but in the general case it will result in a small loss of precision.

Additionally, the TimeSpan.FromSeconds method can overflow when dealing with large values. This can be reproduced even when not averaging:

TimeSpan.FromSeconds(TimeSpan.MaxValue.TotalSeconds)  // will throw an OverflowException

Patrick Hofman made a good point in the comments that just switching from seconds to ticks could also result in overflow, and thus another solution is needed.

Adapting the technique given in this answer , we can compute the average of a collection of TimeSpan values using an arithmetic mean approach, without loss of precision and without overflowing:

public static TimeSpan Mean(this ICollection<TimeSpan> source)
{
    if (source == null)
        throw new ArgumentNullException(nameof(source));

    long mean = 0L;
    long remainder = 0L;
    int n = source.Count;
    foreach (var item in source)
    {
        long ticks = item.Ticks;
        mean += ticks / n;
        remainder += ticks % n;
        mean += remainder / n;
        remainder %= n;
    }

    return TimeSpan.FromTicks(mean);
}

Use it similarly to other extension methods, such as:

TimeSpan average = mylistoftimespans.Mean();