Bug fix release!

This release fixes a "bug" in the permutation-corrected clustering (fingerprint) computations.

In the previous implementation, if a given feature dimension had the same value across all words in the list, the permutation-corrected clustering score would come out to 1. The code has now been refactored so that this "edge case" now results in a corrected clustering score of 0.5 (i.e., "chance"), which seems to make more sense.

Details

This bugfix entailed making two changes:

When ranking distances along some feature dimension, take into account the possibility that some feature values may be equal
When computing the percentile rank of the corrected score (within the distribution of shuffled scores), instead of computing the proportion of shuffled scores that were strictly less than the observed score, we now compute the mean "point values" as follows:
- shuffled scores that are less than the observed score get 1 point
- shuffled scores that are greater than the observed score get 0 points
- shuffled scores that are equal to the observed score get 0.5 points

Full Changelog: v0.2.1...v0.2.2

Assets 2

03 Jun 17:13

paxtonfitzpatrick

v0.2.1

6c847a4

v0.2.1 (June 2021)

support decoding audio files longer than 60s
replace deprecated/removed pandas .get_values() and .as_matrix() methods with .to_numpy()
- bump minimum required pandas version to account for this
add a module-level __version__ attribute (PEP 396)
drop support for Python 2.7 & 3.5, add support for Python 3.7, 3.8, 3.9
- add corresponding versions to CI test matrix

Assets 3

05 Sep 22:12

andrewheusser

v0.2.0

3f16b05

v0.2.0 (February 2018)

Enhancements:

accuracy, spc, pfr/pnr and lag-crp all extended to support high-dimensional stimuli (like movies):
2 new matching functions (+ classic approach) to create a recall matrix / run analyses:
- exact - the presented and recalled stimulus matching exactly is chosen (this is the classic approach)
- best - the presented stimulus that most closely matches a the recalled stimulus is chosen
- smooth - a weighted average of the stimuli are chosen, where the weights are derived by the similarity between the recalled item and each presented item.
- support for a variety of distance functions (to compare presented/recalled stimuli) with a new argument, distance (anything supported by scipy.spatial.distance.cdist).
eggs have been revamped! Changes:
- item and features are now combined in a dataframe of dictionaries replacing egg.pres (previously just the items with a separate 'features' field)
- features for recall items now supported. egg.rec is now a dataframe of dictionaries (previously egg.rec was a dataframe of items`
- added egg.get_pres_items which returns a dataframe of presented items
- added egg.get_pres_features which returns a dataframe of features of the presented items
- added egg.get_rec_items which returns a dataframe of recalled items
- added egg.get_rec_features which returns a dataframe of features of the recalled items
- added support to pass a recall matrix (recmat) instead of pres and rec when constructing an Egg
- save format has been changed from pickles to hdf5 for better cross-version compatibility
- Analysis can be performed by calling egg.analyze() (does the same thing as quail.analyze(egg)
a new class FriedEgg containing the result of an analysis of an egg. Includes plot and save methods.
a new function load for loading Eggs, FriedEggs and example data (old load function renamed loadEL)