3623
93%

Ran 12 Jan 2019 09:21PM UTC

Jobs 6

Files 20

Run time 8min

Badge

Embed ▾

pending completion

Build # 3623

Build Type

push

travis-ci

Committed by

web-flow

Commit Message

Add optimized methods to compute grouped reductions (#1651)

Reductions can be computed much more efficiently by directly accumulating values
into the final column vector. Detect common reduction functions, and use optimized
methods for them. For sum and mean, this has the drawback of reducing accuracy
since pairwise summation cannot be used.

This requires two related changes: GroupedDataFrame needs to store the group indices,
and no group indices are empty. The latter implies changing the handling of
CategoricalArray to remove empty groups earlier, which actually makes the code cleaner.
Also reorganize code to process one pair at a time, which makes it simpler and allows
using the optimized methods even when mixed with custom functions.

Coverage Stats

1935 of 2453 relevant lines covered (78.88%)

7977.97 hits per line

Jobs

ID	Job ID	Ran	Coverage
1	3623.1	12 Jan 2019 09:21PM UTC	66.69	Travis Job 3623.1
2	3623.2	12 Jan 2019 09:21PM UTC	67.02	Travis Job 3623.2
3	3623.3	12 Jan 2019 09:22PM UTC	66.69	Travis Job 3623.3
4	3623.4	12 Jan 2019 09:23PM UTC	67.02	Travis Job 3623.4
5	3623.5	12 Jan 2019 09:21PM UTC	79.05	Travis Job 3623.5
6	3623.6	12 Jan 2019 09:30PM UTC	79.05	Travis Job 3623.6

Source Files on build 3623

Detailed source file information is not available for this build.

JuliaData / DataFrames.jl / 3623
93%

README BADGES
x

Markdown

Textile

RDoc

HTML

Rst

Jobs

Source Files on build 3623

JuliaData / DataFrames.jl / 3623 93%

README BADGES x

Markdown

Textile

RDoc

HTML

Rst

Jobs

Source Files on build 3623

JuliaData / DataFrames.jl / 3623
93%

README BADGES
x