TSOMpy

Online measurement techniques, e.g., moving averages (MAs), calculate time-dependent statistics and are often used in adaptive systems. We proposed a framework to compare multiple MA methods, extended MAs to moving histograms (MHs) and time-dependent rate measurment (TDRM) methods in the paper.

TSOMpy is a Python library for online measurement of time series implementing all moving averages (MAs), moving histograms (MHs), and time-dependent rate measurement (TDRM) concepts of the paper.

Features

TSOMpy offers the following features:

Qt5-based GUI

TSOMpy includes an intuitive GUI for simplified use. The GUI offers a project model for working on multiple analyses in parallel, a plot model for generating multiple plots for a specific time series and an analysis model to support various evaluation methods on a specific plot.

Generation and import of time series

TSOMpy allows to import or generate time series using a GUI wizard. Online measurement concepts can be applied to two different types of time series:

• Equally-spaced time series
  • Vector of samples
    • Import a vector of samples (from a CSV file)
    • Generate a vector of samples according to a distribution
    • Set a fixed sample size
  • Vector of time instants
    • Samples are assigned with time indizes 0, 1, 2, ..., n per default
    • Optional: set the starting point and the spacing interval
• Unequally-spaced time series
  • Vector of samples (as above)
  • Vector of time instants
    • Import a vector of time instants (from a CSV file)
    • Generate a vector of time instants according to a distribution

Application of Online Measurement Methods

All online measurement concepts are implemented in an object-oriented class hierarchy. MA and MH concepts inherit from their base class, TDRM interit from their base class and specific MA classes.

MAs (for evenly-spaced time series)

Class name	Full method's name
CumMean	Cumulative Mean (3.1.2)
WMA	Window MA (3.1.3)
DWMA	Disjoint Windows MA (3.1.4)
UEMA	Unbiased Exponential MA (3.1.5)
EMA	Exponential MA (3.1.6)

Time-dependent MAs (for unevenly-spaced time series)

Class name	Full method's name
TWMA	Time Window MA (3.2.2)
DTWMA	Disjoint Time Windows MA (3.2.3)
UTEMA	Unbiased Time-Exponential MA (3.2.4)
TEMA	Time-Exponential MA (3.2.5)

MHs (for evenly-spaced time series)

Class name	Full method's name
MH_CumMean	MH with CumMean (5.1)
MH_WMA	MH with WMA (not considered in the paper)
MH_DWMA	MH with DWMA (not considered in the paper)
MH_UEMA	MH with UEMA (5.1)
MH_EMA	MH with EMA (not considered in the paper)

Time-dependent MHs (for unevenly-spaced time series)

Class name	Full method's name
TDMH_TWMA	TDMH with TWMA (not considered in the paper)
TDMH_DTWMA	TDMH with DTWMA (not considered in the paper)
TDMH_UTEMA	TDMH with UTEMA (5.2)
TDMH_TEMA	TDMH with TEMA (not considered in the paper)

Time-dependent RMs (for unevenly-spaced time series)

Class name	Full method's name
TDRM_TWMA	TDRM with Time Window MA (6.1.2)
TDRM_DTWMA	TDRM with DTWMA (6.1.3)
TDRM_DTWMA_UEMA	TDRM with DTWMA and UEMA (6.1.4)
TDRM_UTEMA	TDRM with UTEMA (6.1.5)
TDRM_UTEMA_CPA	TDRM with UTEMA and Continuous Packet Arrivals (6.1.6)