FlexFringe-python

Python wrapper for flexfringe

Installation

pip install git+https://github.com/tudelft-cda-lab/FlexFringe-python.git

Also be sure to download flexfringe itself. You will need to point the python wrapper to the binary, or put it in your PATH.

If you want to use flexfringe.show() to display the learned models, you also need to have graphviz installed and available.

Usage

Abbadingo formatted input:

    from flexfringe import FlexFringe

    tracefile = "/path/to/your/tracefile"

    flexfringe = FlexFringe(
        flexfringe_path="/path/to/flexfringe",
        heuristic_name="alergia",
        data_name="alergia_data"
    )

    # Learn a state machine
    flexfringe.fit(tracefile)

    # Display the learned state machine
    flexfringe.show()

    # Use state machine to predict likelihoods
    df = flexfringe.predict(tracefile)

    print(df.head())

prints:

       abbadingo type abbadingo length  ... mean scores min score
row nr                                  ...                      
0                   1               10  ...    -2.00281  -2.80362
1                   1               14  ...    -2.57718  -2.80362
2                   1               27  ...    -2.39332  -3.69244
3                   1               25  ...    -2.32146  -3.62624
4                   1                7  ...    -2.15263  -3.07357

[5 rows x 8 columns]

Process finished with exit code 0

Csv input:

It is also possible to use csv files or even dataframes as input:

import pandas as pd
from flexfringe import FlexFringe

tracefile = "/path/to/tracefile.csv"

df_tracefile = pd.read_csv(tracefile)
df_tracefile = df_tracefile.rename(columns={"State": "symb"})

flexfringe = FlexFringe(
    flexfringe_path="/path/to/flexfringe",
    heuristic_name="alergia",
    data_name="alergia_data",
    slidingwindow=1,
    swsize=10,
)

# Learn a state machine
flexfringe.fit(df_tracefile,
               sinkson=1,
               sinkcount=100)

# Use state machine to predict likelihoods
df = flexfringe.predict(df_tracefile)

print(df.head())

note the line df_tracefile = df_tracefile.rename(columns={"State": "symb"})

You can put special prefixes in column names so flexfringe knows what to do with them:

prefix	function
id	trace identifier
type	trace type
symb	symbol
eval	evaluation function data
attr	symbol attribute
tattr	trace attribute

To use a sliding window on the symbols in a csv file, you just need to mark one or more columns as symb and flexfringe will handle the rest for you. Also see the slidingwindow=1 and swsize=10 parameters.