Overview

Extracting extreme events from daily time series data.

Such extreme events are defined as the values exceeding a certain threshold, and lasting for a certain duration.

Extremes events are identified by attributes such as:

- extreme_start_time
- extreme_end_time
- extreme_duration
- sign_start_time
- sign_end_time
- sign_duration
- max
- min
- mean

The algorithm has some robust features, such as:

(P: positive sign, N: negative sign, E: extreme value)

- single opposite valued data are ignored. e.g, P,P,P,N,P,P,P -> P,P,P,P,P,P
- Multiple extreme events within one sign event are considered as one event. e.g, P,P,P,E,P,E,P,P,P is considered as one event.

Installation

pip install eventextreme