julia> TimeFrame("250ms")
TimeFrames.MonthBegin(250 months, Begin::TimeFrames.Boundary = 1)

it should be millisecond not Month Start

similar to femtotrader/TimeSeriesIO.jl#17

Hello,

I don't know if it's a good practice for Julia code but I think
tf macro should be named TF instead.

It have been added by @iblis17
in 378b796

So it should be possible to do:

tf = TF"4H"

Your opinion?

method(s) matching *(::Int64, ::TimeFrame) are required

julia> DateTime(2010,1,1) + TimeFrame("MS")
2010-02-01T00:00:00

julia> DateTime(2010,1,1) + TimeFrame("3MS")
2010-04-01T00:00:00

julia> DateTime(2010,1,1) + 3*TimeFrame("MS")
ERROR: MethodError: no method matching *(::Int64, ::TimeFrames.MonthBegin)
Closest candidates are:
  *(::Any, ::Any, ::Any, ::Any...) at operators.jl:424
  *(::Real, ::Complex{Bool}) at complex.jl:254
  *(::Real, ::Complex) at complex.jl:266
  ...

PS: MS = monthly start

It should be possible to use StepRange with a TimeFrame.

julia> DateTime(2010,1,1):TimeFrame("5T"):DateTime(2012,1,1)
ERROR: MethodError: no method matching zero(::TimeFrames.TimePeriodFrame{Base.Dates.Minute})
Closest candidates are:
  zero(::Type{Base.LibGit2.Oid}) at libgit2/oid.jl:88
  zero(::Type{Base.Pkg.Resolve.VersionWeights.VWPreBuildItem}) at pkg/resolve/versionweight.jl:80
  zero(::Type{Base.Pkg.Resolve.VersionWeights.VWPreBuild}) at pkg/resolve/versionweight.jl:120
  ...
 in steprange_last(::DateTime, ::TimeFrames.TimePeriodFrame{Base.Dates.Minute}, ::DateTime) at ./range.jl:27
 in colon(::DateTime, ::TimeFrames.TimePeriodFrame{Base.Dates.Minute}, ::DateTime) at ./range.jl:118

We might also support non regulary spaced TimeFrames

like http://pandas.pydata.org/pandas-docs/version/0.19.1/timeseries.html#dateoffset-objects

see https://github.com/pandas-dev/pandas/blob/cebc70cc277462176edea73bb71eaad9f83e9f47/pandas/tseries/offsets.py
and https://github.com/pandas-dev/pandas/blob/cebc70cc277462176edea73bb71eaad9f83e9f47/pandas/tseries/tests/test_offsets.py

was closed by #45 #47

see dysonance/Temporal.jl#2 (comment)

Weekly TimeFrame should support firstdayofweek different from Monday

See also: JuliaLang/julia#19208 and JuliaLang/julia#19212

I'm ok to move this repository to @JuliaStats if it helps to have a better TimeSeries implementation for Julia (see discussion )

pinging @ararslan @iblis17

Similar issue with TimeSeriesResampler
femtotrader/TimeSeriesResampler.jl#19

same as femtotrader/DukascopyTicksReader.jl#15

tf = TimeFrame("15MIN")
TimeFrames.TimePeriodFrame{Base.Dates.Minute}(15 minutes,Begin::TimeFrames.Boundary = 1)

but

tf = TimeFrame("15Min")

raises

ERROR: Can't parse '15Min' to TimeFrame
 in TimeFrames.TimeFrame(::String) at /Users/femto/.julia/v0.5/TimeFrames/src/TimeFrames.jl:78

Related issue:
#3

JuliaStats/TimeSeries.jl#264
JuliaLang/julia#18024

TimeFrame could be supported
and we could define a range with not regularly spaced events

Example:

In [1]: import pandas as pd
In [2]: from pandas.tseries.frequencies import to_offset
In [3]: from pandas.tseries.offsets import MonthEnd
In [2]: pd.date_range("2010-01-01", "2020-01-01", freq="MS")
Out[2]:
DatetimeIndex(['2010-01-01', '2010-02-01', '2010-03-01', '2010-04-01',
               '2010-05-01', '2010-06-01', '2010-07-01', '2010-08-01',
               '2010-09-01', '2010-10-01',
               ...
               '2019-04-01', '2019-05-01', '2019-06-01', '2019-07-01',
               '2019-08-01', '2019-09-01', '2019-10-01', '2019-11-01',
               '2019-12-01', '2020-01-01'],
              dtype='datetime64[ns]', length=121, freq='MS')

In [3]: pd.date_range("2010-01-01", "2020-01-01", freq="M")
Out[3]:
DatetimeIndex(['2010-01-31', '2010-02-28', '2010-03-31', '2010-04-30',
               '2010-05-31', '2010-06-30', '2010-07-31', '2010-08-31',
               '2010-09-30', '2010-10-31',
               ...
               '2019-03-31', '2019-04-30', '2019-05-31', '2019-06-30',
               '2019-07-31', '2019-08-31', '2019-09-30', '2019-10-31',
               '2019-11-30', '2019-12-31'],
              dtype='datetime64[ns]', length=120, freq='M')

In [4]: pd.date_range("2010-01-01 20:00:00", "2010-01-04 00:00:00", freq="D")
Out[4]:
DatetimeIndex(['2010-01-01 20:00:00', '2010-01-02 20:00:00',
               '2010-01-03 20:00:00'],
              dtype='datetime64[ns]', freq='D')

In [5]: pd.date_range("2010-01-01", "2010-01-04", freq="D")
Out[5]: DatetimeIndex(['2010-01-01', '2010-01-02', '2010-01-03', '2010-01-04'], dtype='datetime64[ns]', freq='D')

julia> using TimeFrames
julia> using TimeFrames: Begin, End

this is correct:

julia> apply(Monthly(boundary=Begin), Date(2010, 2, 20))
2010-02-01

but

julia> apply(Monthly(boundary=End), Date(2010, 2, 20))
2010-03-01

we should get 2010-02-28

With

apply(Monthly(boundary=End), DateTime(2010, 2, 20))

we should get 2010-02-28T23:59:59.999

This package is currently tested against Julia 0.7 version.
It shouldn't be a problem with 1.0... but it should be done

Maybe a tonext method should be implemented similar to
https://docs.julialang.org/en/stable/manual/dates/#Adjuster-Functions-1https://docs.julialang.org/en/stable/stdlib/dates/#Base.Dates.tonext-Tuple{Function,Base.Dates.TimeType}

tonext(tf::TimeFrame, dt::Dates.TimeType) = apply(tf, dt) + tf

Goal

Define TimeFrame type for a better resampling API for timeseries (TimeSeries.jl, JuliaTS.jl)

Related issues

• JuliaStats/TimeSeries.jl#257
• tanmaykm/JuliaTS.jl#9

StackOverflow question

• http://stackoverflow.com/questions/20816813/resampling-a-dataframe-to-hourly-15min-and-5min-periods-in-julia

from b8dea88#commitcomment-23009515

pinging @tkelman

M TimeFrame shortcut means monthly (end of month)

julia> TimeFrame("M")
TimeFrames.MonthEnd(1 month, End::TimeFrames.Boundary = 2)

julia> DateTime(2010,1,1) + TimeFrame("M")
2010-02-01T00:00:00

We should expect

julia> DateTime(2010,1,1) + TimeFrame("M")
2010-01-31T23:59:59.999

(considering that smallest resolution of a DateTime is 1 millisecond)

Same issue with adding timeframe with boundary End to a Date

julia> Date(2010,1,1) + TimeFrame("M")
2010-02-01

but we should expect

julia> Date(2010,1,1) + TimeFrame("M")
2010-01-31

(considering that smallest resolution of a Date is one day

julia> using TimeFrames

julia> using TimeFrames: End, Begin

julia> Yearly(boundary=End) == Yearly(boundary=Begin)
true

it should be false

range docstring is

  range(start, [step], length)

  Construct a range by length, given a starting value and optional step (defaults to 1).

TimeFrames range is using stop as third parameter not length...

So this is not consistent with Base

Related issue #5 #6

There is now TimePeriodFrame and DatePeriodFrame

both have a field named time_period.

This field should have a more generic name such as period

It should be possible to do:

tf = TimeFrame(Dates.Minute(15))

• Add doctrings to code
• Use Documenter.jl

Similar to https://github.com/femtotrader/DataReaders.jl/issues/25

According http://docs.julialang.org/en/release-0.5/manual/types/#immutable-composite-types
it may be more efficient to have TimeFrame immutable.

same as femtotrader/TimeSeriesResampler.jl#8

JuliaLang/METADATA.jl#10186

TimeFrame("1W")

works as expected but

TimeFrame("W")

raises

ERROR: ArgumentError: invalid base: base must be 2 ≤ base ≤ 62, got 0
 in tryparse_internal(::Type{Int64}, ::SubString{String}, ::Int64, ::Int64, ::Int64, ::Bool) at ./parse.jl:66
 in parse(::Type{Int64}, ::SubString{String}) at ./parse.jl:152
 in parse(::Type{Int64}, ::SubString{String}) at /Applications/Julia-0.5.app/Contents/Resources/julia/lib/julia/sys.dylib:?
 in TimeFrames.TimeFrame(::String) at /Users/femto/.julia/v0.5/TimeFrames/src/TimeFrames.jl:7

see femtotrader/TimeSeriesResampler.jl#3

PS:
see also Python Pandas

• http://pandas.pydata.org/pandas-docs/stable/timeseries.html#offset-aliases
• http://pandas.pydata.org/pandas-docs/stable/timeseries.html#resampling
• http://pandas.pydata.org/pandas-docs/stable/whatsnew.html#resample-api
• https://github.com/pandas-dev/pandas/blob/cebc70cc277462176edea73bb71eaad9f83e9f47/pandas/tseries/offsets.py
• https://github.com/pandas-dev/pandas/blob/cebc70cc277462176edea73bb71eaad9f83e9f47/pandas/tseries/tests/test_offsets.py

In [1]: from pandas.tseries.frequencies import to_offset
In [2]: from pandas.tseries.offsets import YearBegin, YearEnd, Hour

In [3]: YearBegin().apply(datetime.datetime(2016, 7, 20, 13, 14, 35, 245))
Out[3]: Timestamp('2017-01-01 13:14:35.000245')

In [4]: YearEnd().apply(datetime.datetime(2016, 7, 20, 13, 14, 35, 245))
Out[4]: Timestamp('2016-12-31 13:14:35.000245')

In [5]: to_offset('AS').apply(datetime.datetime(2016, 7, 20, 13, 14, 35, 245))
Out[5]: Timestamp('2017-01-01 13:14:35.000245')

In [6]: to_offset('A').apply(datetime.datetime(2016, 7, 20, 13, 14, 35, 245))
Out[6]: Timestamp('2016-12-31 13:14:35.000245')

there is

typealias DateOrDateTime Union{Date,DateTime}

maybe we should use TimeType and

using Base.Dates: TimeType

We should provide a syntax such as

tf=TimeFrame("M", boundary=End)

dt = DateTime(2010, 1, 1, 0, 0)
tonext(TimeFrame("2H"), dt)

returns

2010-01-01T00:00:00

it should return

2010-01-01T02:00:00

Microsecond, Nanosecond time type is required

see

• https://github.com/JuliaLang/julia/pull/12274
• https://github.com/JuliaLang/julia/issues/12140
• https://groups.google.com/forum/#!topic/julia-users/IvUuQamC5D0
• https://groups.google.com/forum/#!topic/julia-users/R25Mz7nWo7o

Maybe freqstr (like Pandas attribute for DateOffsets)
or simply String

JuliaLang/METADATA.jl#6988

Generating a Project.toml should be considered
Related #49