Inconsistent Handling of Missing Values and Edge Cases Compromises Scientific Results

[anilbey]

Inconsistent Handling of Missing Values and Edge Cases Compromises Scientific Results

Brief:

The current handling of edge cases, especially regarding missing values and certain exceptions in feature extraction, is inconsistent and blurs the distinction between missing and valid scientific values, corrupting our scientific results in downstream analyses, such as group feature extraction and optimisation.

Example problematic missing value representations:

'time_to_first_spike': None,
'time_to_last_spike': array([0.]),
'AP1_amp': array([], dtype=float64),
'decay_time_constant_after_stim': array([-1.])

Solution:

1. Prune Redundant Features:
   Identify and remove unused and untested features, ensuring that documentation is updated accordingly.

2. Standardise Missing Values and Edge Cases Handling:
   Develop solutions for handling missing values and specific exceptions consistently, ensuring clear and unambiguous representation and handling of all data points and states across all active features.

Changes:

• Remove Duplicated code in LibV2-LibV3: Remove copy-pasted code in LibV2-LibV3 #314
• Removal of Redundant Features in Libraries: Removal of Redundant Features in Libraries #minor #322

[anilbey]

This issue is resolved after merging 322, 323, 329 and 334.