wassname/catalyst
1
0
Fork 0
mirror of https://github.com/wassname/catalyst.git synced 2026-07-02 18:46:32 +08:00
Code Issues Packages Projects Releases Wiki Activity
4,516 Commits 16 Branches 39 Tags
f77892240a1024e5e0b51b8eebb3aaec01a98653
Commit Graph

4516 Commits

This Branch
This Branch
All Branches
Author SHA1 Message Date
Scott Sanderson f77892240a MAINT: Use specific versions in appveyor.yml. 2016-09-20 17:12:08 -04:00
Scott Sanderson ddef107fa1 BLD: Downgrade to scipy 0.17. 
Anaconda doesn't have windows builds for scipy 0.18 (nor does
conda-forge.)
 2016-09-20 17:12:08 -04:00
Scott Sanderson 46cf54b180 MAINT: Remove outdated compat code. 2016-09-20 17:12:07 -04:00
Scott Sanderson c6dc1db1f2 BLD: Update appveyor.yml for new pandas/numpy. 2016-09-20 17:12:07 -04:00
Scott Sanderson 6aeba11176 STY: Fix flake8 failures. 2016-09-20 17:12:07 -04:00
Scott Sanderson ebbd79c7db MAINT: Put scipy back in travis reqs. 2016-09-20 17:12:07 -04:00
Scott Sanderson 9e517691bb BLD: Remove old numpy/pandas versions from travis. 2016-09-20 17:12:07 -04:00
Scott Sanderson 48a44f321a DOC: Note where cleanup happens. 2016-09-20 17:12:07 -04:00
Scott Sanderson 4ba064dae7 MAINT: Don't make datetime64 from tz-aware Timestamp. 
It's slow and deprecated.
 2016-09-20 17:12:07 -04:00
Scott Sanderson f24db9c30c DOC: Typo in comment. 2016-09-20 17:12:07 -04:00
Scott Sanderson b639555404 MAINT: Use errors='coerce'. 
coerce=True is deprecated.
 2016-09-20 17:12:07 -04:00
Scott Sanderson 493e18252d MAINT: Temporarily ignore pandas warnings in categoricals. 
Pandas 0.18 doesn't like having null-ish values in categoricals.  Fixing
this properly requires re-thinking the semantics for missing_value on
pipeline terms, so we're punting on that until after we've upgraded to
0.18.
 2016-09-20 17:12:07 -04:00
Scott Sanderson a9c02935c6 Revert "MAINT: Remove support for custom string Column missing values." 
This reverts commit 1b1e842e2339d6d0ee40cdfe34dcd27b4e4a7c0c.
 2016-09-20 17:12:07 -04:00
Scott Sanderson ed365dc5fe MAINT: Remove support for custom string Column missing values. 
Pandas 0.18 deprecated passing "null-ish" values to pd.categorical.  The
expectation, instead, is that you use categorical's native support for
missing data, which means the user will always get NaN's for missing
entries of the categorical.

A follow-up to this change should probably drop support for custom
missing values entirely and to use LabelArray/categorical for integer
data.
 2016-09-20 17:12:07 -04:00
Scott Sanderson da8ed8919e MAINT: Pandas compat for rolling_*. 2016-09-20 17:12:07 -04:00
Scott Sanderson 52166e9958 MAINT: Pass float to np.full explicitly. 2016-09-20 17:12:07 -04:00
Scott Sanderson 4c3c2fd13c MAINT: Don't use convert_objects(). 
It's not necessary, and it's deprecated.
 2016-09-20 17:12:07 -04:00
Scott Sanderson a1273cd669 MAINT: Fix warnings from numpy labelarray methods. 2016-09-20 17:12:07 -04:00
Scott Sanderson af5f4be17c MAINT: Fix warnings from numpy on NaT comparison. 2016-09-20 17:12:07 -04:00
Scott Sanderson 9bd6cab115 MAINT: Pass float explicitly. 2016-09-20 17:12:07 -04:00
Scott Sanderson b9c2e8c1c6 MAINT: Use sort_values instead of sort(). 
Sort is deprecated.
 2016-09-20 17:12:07 -04:00
Scott Sanderson 9aa866e434 MAINT: Use sort_values() instead of sort(). 
pd.DataFrame.sort() is deprecated.
 2016-09-20 17:12:07 -04:00
Scott Sanderson f30fc28b5c MAINT: Pass float explicitly to numpy. 2016-09-20 17:12:07 -04:00
Scott Sanderson ef88dfdad2 MAINT: Use dataframe.iteritems instead of iterkv. 
iterkv is deprecated.
 2016-09-20 17:12:07 -04:00
Scott Sanderson 74277490d5 BUG: Use arrays for week/month group calculations. 
In pandas 0.18, the behavior of ``nth()`` changed so that Grouper no
longer can be easily used to recover group labels.

Instead of using the built-in grouper behavior, we use a groupby on two
arrays we build ourselves.  This recovers the original behavior, and is
about 2x faster as a bonus.
 2016-09-20 17:12:07 -04:00
Scott Sanderson f8734e8721 MAINT/TEST: Clarify test_events. 
- Refer to ``sessions`` instead of periods.
- Use ``toolz.concat`` instead of an O(N ** 2) sequence of appends.
 2016-09-20 17:12:07 -04:00
Scott Sanderson 259f10a2d9 MAINT: Pass float to np.full explicitly. 2016-09-20 17:12:07 -04:00
Scott Sanderson 434d7c69d3 TEST/MAINT: Silence no_checkpoints warning. 2016-09-20 17:12:07 -04:00
Scott Sanderson 905df55a2d MAINT: Use float in np.full. 2016-09-20 17:12:07 -04:00
Scott Sanderson c4e114c28f MAINT: Pass explicit dtype to np.full. 2016-09-20 17:12:07 -04:00
Scott Sanderson dafe49fcca MAINT: Fix failures/warnings in test_api_shim.py 
- Fixes a warning on indexing with a float that ultimately came from
  pd.Timedelta.total_seconds().  Adds ``timedelta_to_integral_seconds``
  and ``timedelta_to_integral_minutes()`` functions and replaces various
  usages of ``int(delta.total_seconds())`` with them.

- Fixes a warnings triggered in ``_create_daily_stats`` from
  passing tz-aware datetimes to np.datetime64.
 2016-09-20 17:12:07 -04:00
Scott Sanderson e43a3e8b1b MAINT: Use arange instead of rolling_count. 
It's faster, and rolling_count is deprecated.
 2016-09-20 16:24:55 -04:00
Scott Sanderson e33c42b3a9 MAINT: Update fetcher for pandas 0.18. 
- Remove support for usecols=[].
- Use sort_values instead of sort().
- Use errors='coerce' instead of coerce=True.
 2016-09-20 16:24:55 -04:00
Scott Sanderson 61a56492e6 ENH: Make INT_DTYPES_BY_SIZE_BYTES ordered. 2016-09-20 16:24:55 -04:00
Scott Sanderson fc3eac36aa DOC: Update LabelArray docstring. 2016-09-20 16:24:55 -04:00
Scott Sanderson b06ef66f44 DOC: Remove out of date comment. 2016-09-20 16:24:55 -04:00
Scott Sanderson 0d8e99956e MAINT: Fix numpy deprecation warnings. 2016-09-20 16:24:55 -04:00
Scott Sanderson be30c0072d MAINT: Explicitly use float64 in test. 2016-09-20 16:24:54 -04:00
Scott Sanderson d2f0632101 MAINT: Don't use .loc with integers. 2016-09-20 16:24:54 -04:00
Scott Sanderson 29d5854aa1 MAINT: Bump numpy and pandas. 2016-09-20 16:24:54 -04:00
Richard Frank 80227eda57 Merge pull request #1493 from quantopian/enumpycal 
Enumpycal
 2016-09-19 16:27:13 -04:00
Richard Frank ef04fe3a9b PERF: Use empyrical with ndarrays instead of Series 2016-09-19 15:51:06 -04:00
Richard Frank 73a61caaeb MAINT: Removing unused array and unused attribute 2016-09-19 15:19:30 -04:00
Jean Bredeche 657e5f92e7 Merge pull request #1497 from quantopian/you-count-way-too-slowly 
PERF: Be smarter about counting the number of minutes across a contiguous bunch of sessions.
 2016-09-19 14:34:11 -04:00
Jean Bredeche 8048cdcfb0 PERF: Be smarter about counting the number of minutes across a contiguous bunch of sessions. 2016-09-19 13:25:03 -04:00
Andrew Daniels 4a2faa3e13 ENH: Adds truncate method to BcolzMinuteBarWriter (#1499) 2016-09-19 13:02:48 -04:00
Eddie Hebert 72a8316ca7 Merge pull request #1496 from quantopian/revert-load-adjustments 
Revert "Merge pull request #1490 from quantopian/use-load-adjustments…
 2016-09-19 09:38:02 -04:00
Eddie Hebert 6486e64530 Revert "Merge pull request #1490 from quantopian/use-load-adjustments-for-history" 
This reverts commit 86c7635b45, reversing
changes made to c77f2b92df.

Some real world cases hit errors with this change, due to the new offset
logic attempting to create Adjustments with invalid parameters.

Will identify exact conditions that cause this error and add as a test
case before remerging.
 2016-09-19 08:53:31 -04:00
Jean Bredeche e21fa37e22 Merge pull request #1492 from quantopian/cache-minute-to-session-lookups 
PERF: Save up to 80% of the calls to minute_to_session_label
 2016-09-16 16:07:13 -04:00
Jean Bredeche 31cbdde7f0 PERF: Save up to 75% of the calls to minute_to_session_label 
One year NYSE test that buys a lot triggers 492,963 calls to
minute_to_session_label.  Only 98924 ~(390 * 252) make it past the
cache and trigger the heavier computation.
 2016-09-16 15:18:50 -04:00