Construct from null
Data / Value type aliases
Index / Key / Time type aliases
Index / Key / Time type aliases
Data / Value type aliases
An alias for key-value representation.
An alias for time-series index.
An alias for key-value representation.
ndslice-like primitives
Data is any ndslice with only one constraints, data and index lengths should be equal.
ndslice-like primitives
Gets data for the index.
Gets data for the index.
Gets data for the index.
Gets data for the index (extra verbose exception).
Gets data for the index (verbose exception).
Index series is assumed to be sorted.
ndslice-like primitives
This function uses a search with policy sp to find the largest left subrange on which t < key is true for all t. The search schedule and its complexity are documented in std.range.SearchPolicy.
ndslice-like primitives
ndslice-like primitives
Special [] = index-assign operator for index-series. Assigns data from r with index intersection. If a index index in r is not in the index index for this series, then no op-assign will take place. This and r series are assumed to be sorted.
Special [] op= index-op-assign operator for index-series. Op-assigns data from r with index intersection. If a index index in r is not in the index index for this series, then no op-assign will take place. This and r series are assumed to be sorted.
ndslice-like primitives
Tries to get the first value, such that key_i == key.
Tries to get the first value, such that lowerBound <= key_i <= upperBound.
Tries to get the first value, such that lowerBound <= key_i <= upperBound. Updates lowerBound with key_i.
Tries to get the last value, such that lowerBound <= key_i <= upperBound.
Tries to get the last value, such that lowerBound <= key_i <= upperBound. Updates upperBound with key_i.
Tries to get the first value, such that key_i >= key.
Tries to get the first value, such that key_i >= key. Updates key with key_i.
Tries to get the last value, such that key_i <= key.
Tries to get the last value, such that key_i <= key. Updates key with key_i.
This function uses a search with policy sp to find the largest right subrange on which t > key is true for all t. The search schedule and its complexity are documented in std.range.SearchPolicy.
1-dimensional data
auto index = [1, 2, 3, 4]; auto data = [2.1, 3.4, 5.6, 7.8]; auto series = index.series(data); const cseries = series; assert(series.contains(2)); assert( ()@trusted{ return (2 in series) is &data[1]; }() ); assert(!series.contains(5)); assert( ()@trusted{ return (5 in series) is null; }() ); assert(series.lowerBound(2) == series[0 .. 1]); assert(series.upperBound(2) == series[2 .. $]); assert(cseries.lowerBound(2) == cseries[0 .. 1]); assert(cseries.upperBound(2) == cseries[2 .. $]); // slicing type deduction for const / immutable series static assert(is(typeof(series[]) == Series!(int*, double*))); static assert(is(typeof(cseries[]) == Series!(const(int)*, const(double)*))); static assert(is(typeof((cast(immutable) series)[]) == Series!(immutable(int)*, immutable(double)*))); /// slicing auto seriesSlice = series[1 .. $ - 1]; assert(seriesSlice.index == index[1 .. $ - 1]); assert(seriesSlice.data == data[1 .. $ - 1]); static assert(is(typeof(series) == typeof(seriesSlice))); /// indexing assert(series[1] == observation(2, 3.4)); /// range primitives assert(series.length == 4); assert(series.front == observation(1, 2.1)); series.popFront; assert(series.front == observation(2, 3.4)); series.popBackN(10); assert(series.empty);
2-dimensional data
import mir.date: Date; import mir.ndslice.topology: canonical, iota; size_t row_length = 5; auto index = [ Date(2017, 01, 01), Date(2017, 02, 01), Date(2017, 03, 01), Date(2017, 04, 01)]; // 1, 2, 3, 4, 5 // 6, 7, 8, 9, 10 // 11, 12, 13, 14, 15 // 16, 17, 18, 19, 20 auto data = iota!int([index.length, row_length], 1); // canonical and universal ndslices are more flexible then contiguous auto series = index.series(data.canonical); /// slicing auto seriesSlice = series[1 .. $ - 1, 2 .. 4]; assert(seriesSlice.index == index[1 .. $ - 1]); assert(seriesSlice.data == data[1 .. $ - 1, 2 .. 4]); static if (kindOf!(typeof(series.data)) != Contiguous) static assert(is(typeof(series) == typeof(seriesSlice))); /// indexing assert(series[1, 4] == observation(Date(2017, 02, 01), 10)); assert(series[2] == observation(Date(2017, 03, 01), iota!int([row_length], 11))); /// range primitives assert(series.length == 4); assert(series.length!1 == 5); series.popFront!1; assert(series.length!1 == 4);
Construct from null
import mir.series; alias Map = Series!(string*, double*); Map a = null; auto b = Map(null); assert(a.empty); assert(b.empty); auto fun(Map a = null) { }
Plain index series data structure.
*.index[i]/*.key[i]/*.time corresponds to *.data[i]/*.value.
Index is assumed to be sorted. sort can be used to normalise a series.