recreate example data objects work with sklearn>=0.24, upload to Drive to make available
replaced file IDs in hypertools.tools.load to reference new versions
updates to data I/O
HypertoolsIOError class for data loading/saving-related errors
refactor hypertools.tools.load.load and hypertools.datageometry.DataGeomtry.save to use pickle instead of deepdish -- more efficient, and allows dependency on deepdish to be removed, which makes Hypertools compatible with Python 3.9
leave compression argument in hypertools.datageometry.DataGeomtry.save for now, but issue FutureWarning if passed
hypertools.tools.load updates:
add legacy argument to load function to enable loading .geo files from old format
add _load_legacy function for loading legacy-format data using deepdish
rewrite _load_example_data function to load data in new format
rewrite _download_example_data function to write stream to file more safely & automatically clean up in case of error/interruption during write
raise more explicit/clear exceptions on various failure conditions
updates to tests
add tests for Python 3.9
switch from conda to mamba for faster environment setup
run pytest with more verbose output
don't add arm64 architecture tests for now -- test_format_data.test_force_align() fails seemingly due to bug with Ubuntu Xenial version used by Travis
other updates
bump min sklearn version in requirements.txt to 0.24
remove __future__ and builtins imports from code (no longer needed for Python 2/3 compatibility)
remove dependencies on six & future (no longer needed for Python 2/3 compatibility)
fix use of is to compare literals
update readme.md to reflect new requirements
add syntax highlighting to fenced code blocks
update test_format_data.test_force_align to check output shape against new shape of example data
sklearn>=0.24
, upload to Drive to make availablehypertools.tools.load
to reference new versionsHypertoolsIOError
class for data loading/saving-related errorshypertools.tools.load.load
andhypertools.datageometry.DataGeomtry.save
to usepickle
instead ofdeepdish
-- more efficient, and allows dependency ondeepdish
to be removed, which makes Hypertools compatible with Python 3.9compression
argument inhypertools.datageometry.DataGeomtry.save
for now, but issueFutureWarning
if passedhypertools.tools.load
updates:legacy
argument toload
function to enable loading.geo
files from old format_load_legacy
function for loading legacy-format data usingdeepdish
_load_example_data
function to load data in new format_download_example_data
function to write stream to file more safely & automatically clean up in case of error/interruption during writeconda
tomamba
for faster environment setuptest_format_data.test_force_align()
fails seemingly due to bug with Ubuntu Xenial version used by Travissklearn
version inrequirements.txt
to0.24
__future__
andbuiltins
imports from code (no longer needed for Python 2/3 compatibility)six
&future
(no longer needed for Python 2/3 compatibility)is
to compare literalsreadme.md
to reflect new requirementstest_format_data.test_force_align
to check output shape against new shape of example data