SaXing Heart Rate

Loading the data and libraries

First, I load the data logged by my Emfit sleep monitor device, using this library.

# load library for importing JSON data
library(importEmfit)

## Loading required package: jsonlite
## 
## Attaching package: 'jsonlite'
## 
## The following object is masked from 'package:utils':
## 
##     View

# import data
filename <- "~/Dropbox/DataSets/Emfit/device-10-presence-report-31260.json"
data <- importEmfitJSON(filename, timezone="Europe/Ljubljana")

# a summary of the data
summary(data)
tail(data, 100)

# create a heart rate time series
# sampling frequency for heart rate is one measurement per 2 seconds
heartRate <- ts(data$heart_rate, frequency = 1/2)

Load the libraries

# load library TSclust
library(TSclust)

## Loading required package: wmtsa
## Loading required package: splus2R
## Loading required package: ifultools
## Loading required package: MASS
## Loading required package: pdc
## Loading required package: cluster

# load ggplot2 for plotting time series
library(ggplot2)
# load sewave library
library(seewave)

## 
## ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
## Welcome to seewave ! 
## The package is regularly updated, please check for new version [http://rug.mnhn.fr/seewave]
## Thanks to use the right reference when citing seewave in publications
## See citation('seewave')
## ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~

Plot the timeseries

# plot the heart rate using ggplot2
plot <- ggplot(data = data) + geom_point(aes(x=timestamp, y=heart_rate), color="blue") + theme_bw()
plot

plot of chunk unnamed-chunk-2

Testing the TSclust package

# plot the heart rate time series as SaX
# using alphabet size of 4 and word length of 100
# NOTE: the col.ser parameter seems a bit buggy or perhaps I'm not using it correctly
SAX.plot(series=heartRate, w=100, alpha = 4, col.ser=c("black","blue","red"))

## Warning: Series length is not multiple of w (number of frames), plotting
## the dimensionality reduced series

plot of chunk TSclust_variant

# do a PAA on the heart rate time series
PAA(heartRate, w=100)

##   [1] 45.298 65.045 60.690 52.272 68.554 72.016 72.025 71.315 71.054 70.805
##  [11] 52.054 62.359 71.113 71.798 72.526 72.692 71.371 69.404 68.013 68.939
##  [21] 71.723 71.445 69.840 66.279 65.582 64.752 64.844 64.338 65.483 64.785
##  [31] 64.558 65.438 64.301 64.248 58.198 57.345 53.909 59.790 59.539 63.521
##  [41] 63.896 64.733 64.026 64.136 63.327 64.108 62.986 63.208 63.571 63.562
##  [51] 63.472 61.852 64.866 51.725 63.120 63.138 61.953 63.485 58.634 58.429
##  [61] 57.813 57.667 58.169 57.886 57.650 61.224 57.299 57.331 57.747 54.734
##  [71] 58.032 55.492 46.077  8.837 33.818 53.993 58.027 48.581 39.612 55.551
##  [81] 56.775 56.680 58.201 55.942 51.817 51.265 53.147 55.293 56.613 49.766
##  [91] 37.802 51.184 56.043 56.846 33.219 51.029 50.924 52.577 46.108 39.167

# code below does not produce a symbol, but a number
# if they were at least different, I would understand
# but at the moment I believe the function is not working properly...
convert.to.SAX.symbol(PAA(heartRate,w=100), alpha = 4)

##   [1] 4 4 4 4 4 4 4 4 4 4 4 4 4 4 4 4 4 4 4 4 4 4 4 4 4 4 4 4 4 4 4 4 4 4 4
##  [36] 4 4 4 4 4 4 4 4 4 4 4 4 4 4 4 4 4 4 4 4 4 4 4 4 4 4 4 4 4 4 4 4 4 4 4
##  [71] 4 4 4 4 4 4 4 4 4 4 4 4 4 4 4 4 4 4 4 4 4 4 4 4 4 4 4 4 4 4

The plot is nice, the PAA will be useful. The SAX word should not be a number vector… perhaps a bug?

Testing the seewave package

# produce a SaX of the heart rate time series
# using alphabet size of 4 and word lenght 100
SAX( x=heartRate, alphabet_size = 4, PAA_number = 100 )

##   [1] "a" "c" "c" "b" "d" "d" "d" "d" "d" "d" "b" "c" "d" "d" "d" "d" "d"
##  [18] "d" "c" "d" "d" "d" "d" "c" "c" "c" "c" "c" "c" "c" "c" "c" "c" "c"
##  [35] "b" "b" "b" "c" "c" "c" "c" "c" "c" "c" "c" "c" "c" "c" "c" "c" "c"
##  [52] "c" "c" "b" "c" "c" "c" "c" "b" "b" "b" "b" "b" "b" "b" "c" "b" "b"
##  [69] "b" "b" "b" "b" "a" "a" "a" "b" "b" "a" "a" "b" "b" "b" "b" "b" "b"
##  [86] "b" "b" "b" "b" "a" "a" "b" "b" "b" "a" "b" "b" "b" "a" "a"

The SAX word seems to agree with the graph above. Nice!

Conclusion