timescope 0.1.1

Simple Python library written in Rust for analyzing and scoping out patterns in time series data

TimeScope

Simple Python library written in Rust for analyzing and scoping out patterns in time series data. This is a pet project to practice programming in Rust and currently only contains two algorithms:

• ED-Pelt change point detection
• Steady State detection (based on change points)

Install

pip install timescope

Example Use

from timescope import TimeSeriesData, cpd_ed_pelt, ssd_cpd

A = [0, 0, 0, 0, 10, 10, 10, 10, 10, 0, 0, 0, 0]
B = list(range(len(A)))

cps = cpd_ed_pelt(A, 1)
print(f"change points: {cps}")
# change points: [4, 9]

ts = TimeSeriesData(B, A)
print(ts)
# TimeSeriesData(
#   time=[0, 1, ..., 11, 12],
#   data=[0, 0, ..., 0, 0],
#   granularity=1,
#   is_step=false
# )

resampled_ts = ts.equally_spaced_resampling()
print(resampled_ts.slice(3, 6))
# TimeSeriesData(
#   time=[3, 4, 5, 6],
#   data=[0, 10, 10, 10],
#   granularity=1,
#   is_step=false
# )

ss_map = ssd_cpd(resampled_ts, 1, 1.0, -1.0)
print(f"steady state map: {ss_map.data}")
# steady state map: [1.0, 1.0, 1.0, 1.0, 0.0, 1.0, 1.0, 1.0, 1.0, 0.0, 1.0, 1.0, 1.0]

For more comprehensive examples, check the notebooks folder.