FindPeaks: Detect Local Extrema
In addisonklinke/rainflow: Extract cyclic features from compressed time series

Description Usage Arguments Value Author(s) References Examples

View source: R/FindPeaks.R

Compress a vector (or time series) by extracting only the important local extrema (see references for algorithm implemented).

1	FindPeaks(a, R, smooth, window = NULL, thres = NULL, na.pad = FALSE)

`a`	Vector of measured values (can be time series or not).
`R`	Numeric value greater than 1 indicating the threshold for important local extrema. Larger values mean fewer peaks are selected (higher compression ratio). Try starting with 1.1 if you are unsure what value to use.
`smooth`	Logical, should a moving average filter be applied before extracting local extrema? If `TRUE` (recommended), this is implemented with the `rollmean()` function from the `zoo` package.
`window`	Numeric, if `smooth = TRUE`, a value indicating the window of points to be used for moving average filter. Else if `smooth = FALSE`, this can be left as `NULL` (the default).
`thres`	Numeric value indicating that points in `a` below `thres` will be dropped. This occurs before any smoothing or detection of local extrema. If `thres = NULL` (the default), all points will be retained for analysis.
`na.pad`	Logical, should the compressed (`x, y`) vectors be padded with `NA` at the ends to make them the same length as the original data? Useful if you are trying to add a column to a dataframe and need to match the number of existing rows. Then when plotting, you can use `na.rm = TRUE` to plot only the extrema.

Object of class peaks, a list containing:

`indices`	Locations of detected peaks.
`peaks`	y-values for detected peaks.
`data`	Original data passed to this function.
`smooth`	If `smooth = TRUE`, the smoothed data, else `NA`.
`window`	If `smooth = TRUE`, the smoothing window used, else `NA`.
`compr`	Compression ratio (1 minus number of peaks detected / length of original vector).
`corr`	Pearson correlation coefficient between original and compressed vectors.
`diff`	Mean difference between original and compressed vectors.

Addison Klinke, agk38@case.edu

\insertRef

fink_indexing_2004rainflow

# Smooth with moving average filter of 120 minutes (2 hours)
# Extract local extrema for temperature time series with R=1.1 threshold
data(weather)
p <- FindPeaks(weather$temp[1:7000], R = 1.1, smooth = TRUE, window = 120)
summary(p)

# Plot the compressed and original time series
plot(p)