mt3 0.2.0

A simple module for outlier detection thanks to Modified Thompson Tau Test

Installation

mt3 requires Python 3.8+

To install the package run :

pip install mt3

If you are planing to use it with numpy and/or pandas, add optionnal dependencies :

pip install mt3[pandas, numpy] # or pip install mt3[numpy] for numpy only

mt3 will then be capable to deal with numpy.ndarray and pd.Series.

By default mt3 is provided with a table of Student T critical values. Available confidence levels are [0.9, 0.95, 0.975, 0.99, 0.995, 0.999]. To be able to use any confidence level, add scipy optionnal dependency :

pip install mt3[scipy]

Usage

mt3 main function is modified_thompson_tau_test :

from mt3 import modified_thompson_tau_test

sample = [-4, 3, -5, -2, 0, 1, 1000]

# You can use it with a simple list :

modified_thompson_tau_test(sample, 0.99)
# [False, False, False, False, False, False, True]


# But you can also use it with a numpy.ndarray or a pandas.Series
import numpy as np
import pandas as pd

modified_thompson_tau_test(np.array(sample), 0.99)
# [False False False False False False True] (numpy array)

modified_thompson_tau_test(pd.Series(sample), 0.99)
# [False False False False False False True] (numpy array)

# If you have nan values in your array or Series, you can choose to treat
# them as outliers
sample_with_nan = np.array([-4, np.nan, 3, -5, -2, 0, 1, 1000])

modified_thompson_tau_test(sample_with_nan, 0.99, nan_is_outlier=True)
# [False True False False False False False True] (numpy array)