v0.4.1

What's Changed

• ♻️ improve QOL with mypy by @jvdd in #120
• ⬆️ update pycatch22 by @jvdd in #122
• Update dependencies by @jvdd in #123
• Update deps ctd. by @jvdd in #124
• 🐛 check for None before setting n_jobs to nb logical CPUs by @jvdd in #126
• 🙏 not writeable strided for vectorized by @jvdd in #128
• ⬆️ update dependencies by @jvdd in #130

Full Changelog: v0.4.0...v0.4.1

v0.4.0

New features

Now you can utilize the FeatureCollection.calculate method to compute feature based on group ids.

Specifically, 2 arguments wre atted to this FeatureCollection.calculate method:

• group_by_all: creates groups that contains all rows corresponding to the group value
  • Note that this is +/- identical as passing df.groupby(group_by_all) as data to the .calculate method -> (which is now also a valid input for the data argument)
• group_by_consecutive: creates groups that contain consecutive rows for the group value

Note: Both grouped feature extraction approaches ignore NaNs in the group_by column.

Curious? :Look at our verbose example notebook - grouping feature extraction

What's Changed

• 🎍 improving loggers as described in #66 by @jonasvdd in #73
• 🧹 some necessary maintenance by @jvdd in #80
• 🪵 log % duration + output_names for FeatureCollection by @jvdd in #83
• ✨ validate integration with antropy by @jvdd in #88
• 🎉 validate nolds integration by @jvdd in #94
• 〰️ remove isort and use ruff instead by @jvdd in #99
• ⬆️ support Python 3.11 by @jvdd in #87
• 🎉 validate pyentrp integration by @jvdd in #95
• 🐛 support functools.partial by @jvdd in #104
• 👷 build: create codeql.yml by @NielsPraet in #106
• Build/codspeed setup by @NielsPraet in #107
• ⬆️ update antropy dependency + disable py 3.7 tests by @jvdd in #108
• ⬆️ update dependencies by @jvdd in #111
• ✨ feat: Feature extraction with an identifier by @NielsPraet in #109
• ⬆️ soften pandas lock by @jvdd in #115
• 🚀 Python 3.12 support by @jvdd in #116

New Contributors

• @NielsPraet made their first contribution in #106

Full Changelog: v0.3.0...v0.4.0

tsflex v0.3.0

What's Changed

• 🥅 update make_robust by @jvdd in #45
• 🎨 Update examples by @jvdd in #46
• 💨 series_pipeline insert & append is more compose-like by @jonasvdd in #47
• 📌 Update dependencies by @jvdd in #48
• 🙈 minor bug fix in make_robust by @jvdd in #52
• 🚑 Fix windows bug by @jvdd in #53
• 🚑 update critical depencies by @jvdd in #55
• ✨ vectorized feature function support by @jvdd in #58
• [MRG] Remove deprecated closed argument in pd.daterange by @jeroenboeye in #64
• 🐛 fix bug with bound_method + ✨ new integrations by @jvdd in #62
• ♻️ improve output indexing by @jvdd in #68
• 🖍️ improve the make_robust docs by @jonasvdd in #72
• ✨ decouple stride + support setpoints by @jvdd in #74
• ♻️ refactor indexing + ✂️ decouple stride & window + ✨ support segment idxs by @jvdd in #71

Major changes

Updates on the output indexing

✂️ Decoupling of stride & window from FeatureDescriptors

• Both argments are now optional.
• The stride can now also be a list of stride values

✨ Support segment indexes

• Users can now add their start & end segment indexes to the FeatureCollection.calculate method - allowing even more flexible feature extraction 😉

Full Changelog: v0.2.3...v0.3

---

_{DISCLAIMER: this release was already published some months ago (11 Oct, 2022) on pypi.
Our apologies for the late tag + release on GitHub.}

tsflex v0.2.3

❗ See also: tsflex v0.2.2 which is even more 🔥 than this one

New features

💚 Next to the tsfresh integrations, tsflex's feature extraction now fully integrates with seglearn and tsfel ⬇️

from seglearn.feature_functions import base_features
from tsfel.feature_extraction import get_features_by_domain

from tsflex.features import FeatureCollection, MultipleFeatureDescriptors
from tsflex.features.integrations import seglearn_feature_dict_wrapper, tsfel_feature_dict_wrapper
from tsflex.utils.data import load_empatica_data

# Load sequence-indexed data (in this case a time-index)
df_tmp, df_acc = load_empatica_data(['tmp', 'acc'])

# Construct your feature extraction configuration & extract features
fc = FeatureCollection(
    MultipleFeatureDescriptors(
        functions=[
            *seglearn_feature_dict_wrapper(base_features()),
            *tsfel_feature_dict_wrapper(get_features_by_domain('statistical')),
        ],
        series_names=["TMP", "ACC_x", "ACC_y"],
        windows=["5min", "15min"],
        strides="5min"
    )
)

fc.calculate(data=[df_tmp, df_acc], return_df=True)

Changes

🎉 The FeatureCollection.calculcate it's feauture-DataFrame output now has a determenistic column order see - #40

tsflex v0.2.2

New features

• 🔥 Now also supports feature-extraction on numeric-index data (and thus not only time-based data)
• 💚 Seamless integration with tsfresh, check out the example below:

from tsfresh.feature_extraction import MinimalFCParameters; import scipy.stats as ss

from tsflex.features import FeatureCollection, MultipleFeatureDescriptors
from tsflex.features.integrations import tsfresh_settings_wrapper
from tsflex.utils.data import load_empatica_data

# Load sequence-indexed data (in this case a time-index)
df_tmp, df_acc = load_empatica_data(['tmp', 'acc'])

# Construct your feature extraction configuration & extract features
fc = FeatureCollection(
    MultipleFeatureDescriptors(
        functions=tsfresh_settings_wrapper(MinimalFCParameters()) + [ss.skew],
        series_names=["TMP", "ACC_x", "ACC_y"],
        windows=["5min", "15min"],
        strides="5min"
    )
)

fc.calculate(data=[df_tmp, df_acc], return_df=True)

• ⚡ Optimized strided-rolling feature-extraction, see the newly generated benchmark ⬇️

• Added FeatureCollection.reduce() which comes in really handy when feature selection is performed in your machine-learning pipeline
• 🐻 chunk_data() now also supports DataFrame-dicts as input, which can be more convenient when having DataFrames with a lot of columns for which you want to specify the sample-frequencies.
• 🌻 SeriesPipeline is now more compose-like as it now accepts SeriesPipeline instances

Changes

• 🧵 Changed pathos ➡️ multiprocess as multiprocessing back-end
• 🔧 Moved the bound_method argument to FeatureCollection.calculate()
• 📝 Rewrote strided-rolling back-end in a more OO manner (introduced the segmenter module), which complies with our roadmap of providing more segmenting functionality