PMEventropy

Compute in python entropy for process mining describe in Back, C.O., Debois, S. & Slaats, T. Entropy as a Measure of Log Variability. J Data Semant 8, 129–156 (2019). https://doi.org/10.1007/s13740-019-00105-3 (the article)

This project is inspired by https://github.com/backco/eventropy

Installation

pip install pmentropy

Get started

First import the XES file

import pmentropy
logs = pmentropy.read_file("path", flatten=False)

Then compute an entropy

entropy1 = pmentropy.kNN_entropy(logs, k=3, p=2)
entropy2 = pmentropy.global_block_entropy(logs k=3, p=2)

Import with pm4py

with a DataFrame

import pmentropy
from pm4py.read import read_xes

df = read_xes("path")
logs = pmentropy.read_DataFrame(df, flatten=False)

entropy1 = pmentropy.kNN_entropy(logs, k=3, p=2)

trace by trace if you already read by stream

import pmentropy
from pm4py.streaming.importer.xes import importer as xes_importer

stream = xes_importer.apply("path", variant=xes_importer.xes_trace_stream)
next_trace, logs = pmentropy.read_trace_by_trace(flatten=False)
for trace in stream:
    next_trace(trace)

entropy1 = pmentropy.kNN_entropy(logs, k=3, p=2)

Documentation

Parse file

• read_file(file_path: str, flatten=False)
• read_trace_by_trace(flatten=False)

Entropy

• trace_entropy(logs)
• prefix_entropy(logs)
• unique_trace(logs)
• k_block_entropy(logs, k: int)
• global_block_entropy(logs)
• kL_entropy(logs, p: int)
• kNN_entropy(logs, k: int, p: int)
• lempel_ziv_entropy_rate(logs)
• k_block_entropy_rate_ratio(logs, c)
• k_block_entropy_rate_diff(logs, c)
• unique_trace(logs)