Generate logically periodic time series data.
Timeseries generates logically periodic time steps in any time zone, and accounts for irregularities like daylight-saving time and leap years. As an example, a 1-month period at the beginning of a month preserves the numeric day and time at each step.
2010-01-01 09:05:00
2010-02-01 09:05:00
2010-03-01 09:05:00
An end-of-month series can also be generated; the series automatically maps non-existent times (ex: Feb 31, 02:00 across a DST transition, etc) to their logical equivalents.
2010-01-31 09:05:00
2010-02-28 09:05:00
2010-03-31 09:05:00
Timeseries provides methods to work with these series too, for example to collate data at each time step into intervals.
From the command line:
# start, n_steps, period
$ timeseries -s 2010-01-31 -n 3 -p 1month
2010-01-31 00:00:00 UTC
2010-02-28 00:00:00 UTC
2010-03-31 00:00:00 UTC
# start, stop, period, time format
$ timeseries -s "00:00:00" -t "01:00:00" -p 15m -e "%H:%M"
00:00
00:15
00:30
00:45
01:00
In code:
Time.zone = "UTC"
series = Timeseries.new(
:start_time => Time.zone.parse("2010-01-31"),
:n_steps => 3,
:period => {:months => 1}
)
series.map {|time| time.iso8601 }
# => [
# "2010-01-31T00:00:00Z",
# "2010-02-28T00:00:00Z",
# "2010-03-31T00:00:00Z"
# ]
Use Timeseries.create for more flexible initialization (coerces, parses,
solves for unknowns).
series = Timeseries.create(
:start_time => "00:00:00",
:stop_time => "01:00:00",
:period => "15m"
)
series.map {|time| time.strftime("%H:%M") }
# => [
# "00:00",
# "00:15",
# "00:30",
# "00:45",
# "01:00"
# ]
To collate interval-ending intervals from data corresponding to each step:
series.collate([:a, :b, :c, :d, :e], :format => "%H:%M")
# => {
# "00:15" => [:a, :b],
# "00:30" => [:b, :c],
# "00:45" => [:c, :d],
# "01:00" => [:d, :e]
# }
Leap seconds are not supported.
Setup:
bundle install
git submodule init
git submodule update
Tests:
./test/suite