/// <summary>
/// Makes a collection of empty timestamped workflow trace shells, one for each unique "Case ID" in loaded data from an event log.
/// </summary>
/// <param name="ids">A "Case ID" column from the data frame, representation of a loaded event log.</param>
/// <returns>A list of empty timestamped workflow traces, one for each unique "Case ID" in given data.</returns>
private List <TimestampedWorkflowTrace> MakeEmptyTimestampedWfts(Deedle.Series <int, string> ids)
{
List <TimestampedWorkflowTrace> traces = new List <TimestampedWorkflowTrace>();
HashSet <string> uniqueIds = new HashSet <string>(ids.Values);
foreach (var id in uniqueIds)
{
traces.Add(new TimestampedWorkflowTrace("" + id));
}
return(traces);
}
public void MaxDrawdown()
{
var series = new Deedle.Series <DateTime, double>(new []
{
new KeyValuePair <DateTime, double>(new DateTime(2020, 1, 1), 100000),
new KeyValuePair <DateTime, double>(new DateTime(2020, 1, 2), 90000),
new KeyValuePair <DateTime, double>(new DateTime(2020, 1, 3), 100000),
new KeyValuePair <DateTime, double>(new DateTime(2020, 1, 4), 100000),
new KeyValuePair <DateTime, double>(new DateTime(2020, 1, 5), 80000)
});
var collection = DrawdownCollection.GetDrawdownPeriods(series, 1).ToList();
Assert.AreEqual(1, collection.Count);
Assert.AreEqual(0.2, collection.First().Drawdown, 0.0001);
}
public void Stat2StDevBenchmark()
{
var comparer = KeyComparer <int> .Default;
var count = 1_000_000;
var width = 20;
var sm = new SortedMap <int, double>(count, comparer);
sm.Add(0, 0);
for (int i = 2; i <= count; i++)
{
sm.Add(i, i);
}
var ds1 = new Deedle.Series <int, double>(sm.Keys.ToArray(), sm.Values.ToArray());
var sum = 0.0;
for (int r = 0; r < 10; r++)
{
var sum1 = 0.0;
using (Benchmark.Run("Stat2 Online N", count * width))
{
foreach (var stat2 in sm.Stat2(width).Values)
{
sum1 += stat2.StDev;
}
}
Assert.True(sum1 != 0);
var sum2 = 0.0;
using (Benchmark.Run("Stat2 Online Width GE", count * width))
{
foreach (var stat2 in sm.Stat2(width - 1, Lookup.GE).Values)
{
sum2 += stat2.StDev;
}
}
Assert.True(sum2 != 0);
var sum3 = 0.0;
using (Benchmark.Run("Stat2 Window N", count * width))
{
foreach (var stat2 in sm.Window(width).Map(x => x.Stat2()).Values)
{
sum3 += stat2.StDev;
}
}
Assert.True(sum3 != 0);
var sum4 = 0.0;
using (Benchmark.Run("Stat2 Window Width GE", count * width))
{
foreach (var stat2 in sm.Window(width - 1, Lookup.GE).Map(x => x.Stat2()).Values)
{
sum4 += stat2.StDev;
}
}
Assert.True(sum4 != 0);
var sum5 = 0.0;
using (Benchmark.Run("Deedle (online)", count * width))
{
var deedleStDev = Deedle.Stats.movingStdDev(width, ds1);
foreach (var keyValuePair in deedleStDev.Values)
{
sum5 += keyValuePair;
}
}
Assert.True(sum5 != 0);
Assert.True(Math.Abs(sum1 / sum2 - 1) < 0.000001);
Assert.True(Math.Abs(sum1 / sum3 - 1) < 0.000001);
Assert.True(Math.Abs(sum1 / sum4 - 1) < 0.000001);
Assert.True(Math.Abs(sum1 / sum5 - 1) < 0.000001);
sum = 0.0;
using (Benchmark.Run("SortedMap enumeration", count))
{
var cursor = sm.GetEnumerator();
while (cursor.MoveNext())
{
sum += cursor.currentValue;
}
}
Assert.True(sum != 0);
}
Benchmark.Dump($"The window width is {width}. Stat2 MOPS are calculated as a number of calculated values multiplied by width, " +
$"which is equivalent to the total number of cursor moves for Window case. SortedMap line is for reference - it is the " +
$"speed of raw iteration over SM without Stat2/Windows overheads (and not multiplied by width).");
}
