Type: Package
Title: Decomposes a Level Shifted Time Series
Version: 0.1.1
Description: Explains the behavior of a time series by decomposing it into its trend, seasonality and residuals. It is built to perform very well in the presence of significant level shifts. It is designed to play well with any breakpoint algorithm and any smoothing algorithm. Currently defaults to 'lowess' for smoothing and 'strucchange' for breakpoint identification. The package is useful in areas such as trend analysis, time series decomposition, breakpoint identification and anomaly detection.
License: MIT + file LICENSE
URL: https://allen-1242.github.io/StructuralDecompose/
Depends: R (≥ 2.10)
Imports: changepoint, segmented, strucchange
Suggests: knitr, rmarkdown, testthat (≥ 3.0.0)
VignetteBuilder: knitr
Config/testthat/edition: 3
Encoding: UTF-8
LazyData: true
RoxygenNote: 7.2.3
NeedsCompilation: no
Packaged: 2023-02-11 19:18:15 UTC; sunny
Author: Allen Sunny [aut, cre]
Maintainer: Allen Sunny <allensunny1242@gmail.com>
Repository: CRAN
Date/Publication: 2023-02-13 09:30:04 UTC

Automatic Anomaly detection

Description

Automatic Anomaly detection

Usage

AnomalyDetection(
  timeseries,
  frequency = 52,
  conf_level = 1.5,
  breaks,
  window_len = 14
)

Arguments

timeseries

Given time series

frequency

Timeseries frequency, defaults to 12 points

conf_level

Confidence level for Anomaly detection

breaks

breakpoints identified

window_len

Window length for anomaly detection

Value

the list of anomalies in the time series, along with the time series plot

Examples

AnomalyDetection(timeseries = StructuralDecompose::Nile_dataset[,1], breaks = c(4, 50, 80))

AnomalyDetection(timeseries = runif(n = 50, min = 1, max = 10),  breaks = c(4, 20, 30))

Generation of breakpoints

Description

Generation of breakpoints

Usage

BreakPoints(
  timeseries,
  frequency = 52,
  break_algorithm = "strucchange",
  break_level = 0.05
)

Arguments

timeseries

Given time series

frequency

Timeseries frequency, defaults to 12 points

break_algorithm

Breakpoint algorithm to be used. Defaults to strucchange

break_level

Additional parameters for breakpoint algorithm

Value

A list of breakpoints

Examples

BreakPoints(timeseries = seq(100), frequency = 52, break_level = 0.05)
BreakPoints(timeseries = StructuralDecompose::Nile_dataset[,1], frequency = 52)

Minimum level length checks

Description

Minimum level length checks

Usage

LevelCheck(timeseries, level_length = 10, breaks)

Arguments

timeseries

Given time series

level_length

Mean distance between two levels

breaks

breakpoints returned

Value

The series cleaned with the minimum level check

Examples

LevelCheck(timeseries = StructuralDecompose::Nile_dataset[,1], breaks = c(1,4,5))

LevelCheck(timeseries = runif(n = 50, min = 1, max = 10), breaks = c(1,4,5))

Mean level checks

Description

Mean level checks

Usage

MeanCleaning(timeseries, mean_level = 0.5, breaks, frequency = 52)

Arguments

timeseries

Given time series

mean_level

Mean distance between two levels

breaks

breakpoints returned

frequency

Timeseries frequency, defaults to 12 points

Value

The series cleaned with the mean check

Examples

MeanCleaning(timeseries = StructuralDecompose::Nile_dataset[,1], breaks = c(1,4,5), frequency = 1)

MeanCleaning(timeseries = runif(n = 50, min = 1, max = 10), breaks = c(1,4,5), frequency = 12)

Median level checks

Description

Median level checks

Usage

MedianCleaning(timeseries, median_level = 0.5, breaks, frequency = 52)

Arguments

timeseries

Given time series

median_level

Median distance between two levels

breaks

Breaks identified

frequency

Timeseries frequency, defaults to 12 points

Value

The series cleaned with the median check

Examples

MedianCleaning(timeseries = StructuralDecompose::Nile_dataset[,1], breaks = c(1,4,5))

MedianCleaning(timeseries = runif(n = 50, min = 1, max = 10), breaks = c(1,4,5))

Nile River Dataset

Description

Nile River Dataset

Smoothening of the time series

Description

Smoothening of the time series

Usage

Smoothing(timeseries, frequency = 52, smoothening_algorithm = "lowess", breaks)

Arguments

timeseries

Given time series

frequency

Timeseries frequency, defaults to 12 points

smoothening_algorithm

Smoothening algorithm required

breaks

Breakpoints identified by the previous algorithm

lowess

Lowess smoothener

Value

The smoothened time series

Examples

Smoothing(timeseries = StructuralDecompose::Nile_dataset[,1], breaks = c(4, 50, 80))

Smoothing(timeseries = runif(n = 50, min = 1, max = 10), breaks = c(4, 20, 30))

Main decomposition algorithm

Description

Main decomposition algorithm

Usage

StructuralDecompose(
  Data,
  frequency = 12,
  break_algorithm = "strucchange",
  smoothening_algorithm = "lowess",
  break_level = 0.05,
  median_level = 0.5,
  mean_level = 0.5,
  level_length = 12,
  conf_level = 0.5,
  window_len = 12,
  plot = FALSE
)

Arguments

Data

Time series required

frequency

Frequency of the tine series

break_algorithm

breakpoints algorithm used. Defaults to strucchange

smoothening_algorithm

Smoothing algorithm used. Defaults to lowess

break_level

Break level for the breakpoints algorithm

median_level

Average median distance between two level

mean_level

Average mean distance between a group of points near breakpoints

level_length

Minimum number of points required to determine a level

conf_level

Confidence level for Anomaly detection, best to keep this a static value

window_len

Length of the Moving window for Anomaly Detection

plot

True of False indicating if you want the internal plots to be generated

Value

The decomposed time series along with a host of other metrics

Examples

StructuralDecompose(Data = StructuralDecompose::Nile_dataset[,1])

StructuralDecompose(Data = runif(n = 50, min = 1, max = 10))