tsia: a Python package for time series analysis through images

tsia (time series images analysis) is a package to perform time series analysis and diagnostics thanks to imaging techniques.

Installation

Dependencies

tsia requires:

• python (>= 3.7)
• matplotlib (>= 3.3.0)
• networkx (>= 2.5)
• numba (>= 0.48.0)
• numpy (>= 1.17.5)
• python-louvain (>= 0.14)
• pyts (>= 0.11.0)

User installation

If you already have a working installation of the aforementioned modules, you can easily install tsia using pip:

pip install tsia

Implemented features

tsia consists of the following modules:

• markov: this module provides some methods to compute statistics from Markov transition fields as implemented in the pyts package. It also implements functions to map back MTF output to the original time series.

• network_graph: this module provides some methods to compute statistics and several encoding techniques to interpret network graphs built upon MTF.

• plot: this module implements several useful visualization techniques to provide insights into univariate timeseries. This module implement MTF, network graph visualization and several line graphs (vanilla time series, colored time series with different color encodings, quantile discretization overlay...).

• utils: this modules includes some utilities leveraged by the other ones.