FEEED

Feature Extraction for Event Data

Installation
Usage

Installation

Requirements

Python > 3.9
Java

Clone

Clone this repo to your local machine using

git clone git@github.com:lmu-dbs/feeed.git

To directly use meta feature extraction methods via import

pip install feed

Run:

python -c "from feeed.feature_extractor import extract_features; print(extract_features('test_logs/Sepsis.xes'))"

Usage

Feature types

Specific features can be selected refering their feature types:

Feature Type	Features
simple_stats	n_traces, n_unique_traces, ratio_unique_traces_per_trace
trace_length	trace_len_min, trace_len_max, trace_len_mean, trace_len_median, trace_len_mode, trace_len_std, trace_len_variance, trace_len_q1, trace_len_q3, trace_len_iqr, trace_len_geometric_mean, trace_len_geometric_std, trace_len_harmonic_mean, trace_len_skewness, trace_len_kurtosis, trace_len_coefficient_variation, trace_len_entropy, trace_len_hist1, trace_len_hist2, trace_len_hist3, trace_len_hist4, trace_len_hist5, trace_len_hist6, trace_len_hist7, trace_len_hist8, trace_len_hist9, trace_len_hist10, trace_len_skewness_hist, trace_len_kurtosis_hist
trace_variant	ratio_most_common_variant, ratio_top_1_variants, ratio_top_5_variants, ratio_top_10_variants, ratio_top_20_variants, ratio_top_50_variants, ratio_top_75_variants, mean_variant_occurrence, std_variant_occurrence, skewness_variant_occurrence, kurtosis_variant_occurrence
activities	n_unique_activities, activities_min, activities_max, activities_mean, activities_median, activities_std, activities_variance, activities_q1, activities_q3, activities_iqr, activities_skewness, activities_kurtosis
start_activities	n_unique_start_activities, start_activities_min, start_activities_max, start_activities_mean, start_activities_median, start_activities_std, start_activities_variance, start_activities_q1, start_activities_q3, start_activities_iqr, start_activities_skewness, start_activities_kurtosis
end_activities	n_unique_end_activities, end_activities_min, end_activities_max, end_activities_mean, end_activities_median, end_activities_std, end_activities_variance, end_activities_q1, end_activities_q3, end_activities_iqr, end_activities_skewness, end_activities_kurtosis
entropies	entropy_trace, entropy_prefix, entropy_global_block, entropy_lempel_ziv, entropy_k_block_diff_1, entropy_k_block_diff_3, entropy_k_block_diff_5, entropy_k_block_ratio_1, entropy_k_block_ratio_3, entropy_k_block_ratio_5, entropy_knn_3, entropy_knn_5, entropy_knn_7
complexity	variant_entropy, normalized_variant_entropy, sequence_entropy, normalized_sequence_entropy, sequence_entropy_linear_forgetting, normalized_sequence_entropy_linear_forgetting, sequence_entropy_exponential_forgetting, normalized_sequence_entropy_exponential_forgetting

Examples

Example 1:

Pass sublist ['trace_variant', 'start_activities'] to get a list of values for the features 'trace_variant' and 'start_activities' only

from feeed.feature_extractor import extract_features

features = extract_features("test_logs/Sepsis.xes", ['trace_variant', 'start_activities'])

Output should look like:

{
'log': 'Sepsis'
'ratio_most_common_variant': 0.03333333333333333
'ratio_top_1_variants': 0.12
'ratio_top_5_variants': 0.21523809523809523
'ratio_top_10_variants': 0.2742857142857143
'ratio_top_20_variants': 0.35523809523809524
'ratio_top_50_variants': 0.5971428571428572
'ratio_top_75_variants': 0.7980952380952381
'mean_variant_occurrence': 1.2411347517730495
'std_variant_occurrence': 1.7594085182491936
'skewness_variant_occurrence': 13.637101374069475
'kurtosis_variant_occurrence': 217.44268017168216
'n_unique_start_activities': 6
'start_activities_min': 6
'start_activities_max': 995
'start_activities_mean': 175.0
'start_activities_median': 12.0
'start_activities_std': 366.73787187399483
'start_activities_variance': 134496.66666666666
'start_activities_q1': 7.75
'start_activities_q3': 17.0
'start_activities_iqr': 9.25
'start_activities_skewness': 1.7883562472303318
'start_activities_kurtosis': 1.199106773708694
}

Example 2:

Get a full list of all feature values

from feeed.feature_extractor import extract_features

features = extract_features("test_logs/Sepsis.xes")