HMMpa-package: Analysing Accelerometer Data Using Hidden Markov Models
In HMMpa: Analysing Accelerometer Data Using Hidden Markov Models
Description
Details
Note
Author(s)
References
Examples
Description
This package provides functions for analyzing accelerometer outpout data (known as a time-series of (impulse)-counts) to quantify length and intensity of physical activity.
Usually, so called activity ranges are used to classify an activity as “sedentary”, “moderate” and so on. Activity ranges are separated by certain thresholds (cut-off points). The choice of these cut-off points depends on different components like the subjects' age or the type of accelerometer device.
Cut-off point values and defined activity ranges are important input values of the following analyzing tools provided by this package:
1. Cut-off point method (assigns an activity range to a count given its total magintude).
This traditional approach assigns an activity range to each count of the time-series independently of each other given its total magnitude.
2. HMM-based method (assigns an activity range to a count given its underlying PA-level).
This approach uses a stochastic model (the hidden Markov model or HMM) to identify the (Markov dependent) time-series of physical activity states underlying the given time-series of accelerometer counts. In contrast to the cut-off point method, this approach assigns activity ranges to the estimated PA-levels corresponding to the hidden activity states, and not directly to the accelerometer counts.
Details
Package: HMMpa
Type: Package
Version: 1.0.1
Date: 2018-04-20
License: GPL-3
The new procedure for analyzing accelerometer data can be roughly described as follows:
First, a hidden Markov model (HMM) is trained to estimate the number m of hidden physical activity states and the model specific parameters (delta, gamma, distribution_theta). Then, a user-sepcified decoding algorithm decodes the trainded HMM to classify each accelerometer count into the m hidden physical activity states. Finally, the estimated distribution mean values (PA-levels) corresponding to the hidden physical activity states are extracted and the accelerometer counts are assigned by the total magnitudes of their corresponding PA-levels to given physical activity ranges (e.g. "sedentary", "light", "moderate" and "vigorous") by the traditional cut-off point method.
Note
We thank Moritz Hanke for his help in realizing this package.
Author(s)
Vitali Witowski,
Ronja Foraita,
Leibniz Institute for Prevention Research and Epidemiology (BIPS)
Maintainer: Ronja Foraita <foraita@leibniz-bips.de>
References
Baum, L., Petrie, T., Soules, G., Weiss, N. (1970). A maximization technique occurring in the statistical analysis of probabilistic functions of markov chains. The annals of mathematical statistics, vol. 41(1), 164–171.
Brachmann, B. (2011). Hidden-Markov-Modelle fuer Akzelerometerdaten. Diploma Thesis, University Bremen - Bremen Institute for Prevention Research and Social Medicine (BIPS).
Dempster, A., Laird, N., Rubin, D. (1977). Maximum likelihood from incomplete data via the EM algorithm. Journal of the Royal Statistical Society. Series B (Methodological), vol. 39(1), 1–38.
Forney, G.D. (1973). The Viterbi algorithm. Proceeding of the IEE, vol. 61(3), 268–278.
Joe, H., Zhu, R. (2005). Generalized poisson distribution: the property of mixture of poisson and comparison with negative binomial distribution. Biometrical Journal, vol. 47(2), 219–229.
MacDonald, I. L., Zucchini, W. (2009) Hidden Markov Models for Time Series: An Introduction Using R, Boca Raton: Chapman & Hall.
Viterbi, A.J. (1967). Error Bounds for concolutional codes and an asymptotically optimal decoding algorithm. Information Theory, IEEE Transactions on, vol. 13(2), 260–269.
Witowski, V., Foraita, R., Pitsiladis, Y., Pigeot, I., Wirsik, N. (2014) Using hidden Markov models to improve quantifying physical activity in accelerometer data - A simulation study. PLOS ONE. 9(12), e114089.
http://dx.doi.org/10.1371/journal.pone.0114089
Examples
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
################################################################
### Example 1 (traditional approach) ###################
### Solution of the cut-off point method ###################
################################################################
################################################################
### Fictitious activity counts #################################
################################################################
x <- c(1,16,19,34,22,6,3,5,6,3,4,1,4,3,5,7,9,8,11,11,
14,16,13,11,11,10,12,19,23,25,24,23,20,21,22,22,18,7,
5,3,4,3,2,3,4,5,4,2,1,3,4,5,4,5,3,5,6,4,3,6,4,8,9,12,
9,14,17,15,25,23,25,35,29,36,34,36,29,41,42,39,40,43,
37,36,20,20,21,22,23,26,27,28,25,28,24,21,25,21,20,21,
11,18,19,20,21,13,19,18,20,7,18,8,15,17,16,13,10,4,9,
7,8,10,9,11,9,11,10,12,12,5,13,4,6,6,13,8,9,10,13,13,
11,10,5,3,3,4,9,6,8,3,5,3,2,2,1,3,5,11,2,3,5,6,9,8,5,
2,5,3,4,6,4,8,15,12,16,20,18,23,18,19,24,23,24,21,26,
36,38,37,39,45,42,41,37,38,38,35,37,35,31,32,30,20,39,
40,33,32,35,34,36,34,32,33,27,28,25,22,17,18,16,10,9,
5,12,7,8,8,9,19,21,24,20,23,19,17,18,17,22,11,12,3,9,
10,4,5,13,3,5,6,3,5,4,2,5,1,2,4,4,3,2,1)
traditional_cut_off_point_method <- cut_off_point_method(x=x,
cut_points = c(5,15,23),
names_activity_ranges = c("SED","LIG","MOD","VIG"),
bout_lengths = c(1,1,2,4,5,10,11,20,21,60,61,260),
plotting = 1)
################################################################
################################################################
### Examples 2,3 and 4 (new approach) ########
### Solution of the HMM based cut-off point method ########
################################################################
### Demonstrated both in three steps (Example 2) ######
### and condensed in one function (Examples 3 and 4) ######
################################################################
################################################################
### Example 2) Manually in three steps ######################
################################################################
################################################################
### Fictitious activity counts #################################
################################################################
x <- c(1,16,19,34,22,6,3,5,6,3,4,1,4,3,5,7,9,8,11,11,
14,16,13,11,11,10,12,19,23,25,24,23,20,21,22,22,18,7,
5,3,4,3,2,3,4,5,4,2,1,3,4,5,4,5,3,5,6,4,3,6,4,8,9,12,
9,14,17,15,25,23,25,35,29,36,34,36,29,41,42,39,40,43,
37,36,20,20,21,22,23,26,27,28,25,28,24,21,25,21,20,21,
11,18,19,20,21,13,19,18,20,7,18,8,15,17,16,13,10,4,9,
7,8,10,9,11,9,11,10,12,12,5,13,4,6,6,13,8,9,10,13,13,
11,10,5,3,3,4,9,6,8,3,5,3,2,2,1,3,5,11,2,3,5,6,9,8,5,
2,5,3,4,6,4,8,15,12,16,20,18,23,18,19,24,23,24,21,26,
36,38,37,39,45,42,41,37,38,38,35,37,35,31,32,30,20,39,
40,33,32,35,34,36,34,32,33,27,28,25,22,17,18,16,10,9,
5,12,7,8,8,9,19,21,24,20,23,19,17,18,17,22,11,12,3,9,
10,4,5,13,3,5,6,3,5,4,2,5,1,2,4,4,3,2,1)
################################################################
## Step 1: Training of a HMM ##################################
## for the given time-series of counts ################
## ####################################################
## More precisely: training of a poisson ##############
## distribution based hidden Markov model for #########
## number of states m=2,...,6 #########################
## and selection of the model with the most ###########
## plausible m ########################################
################################################################
m_trained_HMM <- HMM_training(x = x,
min_m = 2,
max_m = 6,
distribution_class = "pois")$trained_HMM_with_selected_m
###############################################################
## Step 2: Decoding of the trained HMM for the given ##########
## time-series of accelerometer counts to extract #####
## hidden PA-levels ###################################
###############################################################
hidden_PA_levels <- HMM_decoding(x = x,
m = m_trained_HMM$m,
delta = m_trained_HMM$delta,
gamma = m_trained_HMM$gamma,
distribution_class = m_trained_HMM
$distribution_class,
distribution_theta = m_trained_HMM$
distribution_theta)$decoding_distr_means
###############################################################
## Step 3: Assigning of user-sepcified activity ranges ########
## to the accelerometer counts via the total ##########
## magnitudes of their corresponding ##################
## hidden PA-levels ###################################
## ####################################################
## In this example four activity ranges ###############
## (named as "sedentary", "light", "moderate" #########
## and "vigorous" physical activity) are ##############
## separated by the three cut-points 5, 15 and 23) ####
################################################################
HMM_based_cut_off_point_method <- cut_off_point_method(x = x,
hidden_PA_levels = hidden_PA_levels,
cut_points = c(5,15,23),
names_activity_ranges = c("SED","LIG","MOD","VIG"),
bout_lengths = c(1,1,2,4,5,10,11,20,21,60,61,260),
plotting=1)
###############################################################
## Example 3) In a single function (Step 1-3 of Example 2 ####
## combined in one function) ####
###############################################################
###############################################################
## Fictitious activity counts #################################
###############################################################
x <- c(1,16,19,34,22,6,3,5,6,3,4,1,4,3,5,7,9,8,11,11,
14,16,13,11,11,10,12,19,23,25,24,23,20,21,22,22,18,7,
5,3,4,3,2,3,4,5,4,2,1,3,4,5,4,5,3,5,6,4,3,6,4,8,9,12,
9,14,17,15,25,23,25,35,29,36,34,36,29,41,42,39,40,43,
37,36,20,20,21,22,23,26,27,28,25,28,24,21,25,21,20,21,
11,18,19,20,21,13,19,18,20,7,18,8,15,17,16,13,10,4,9,
7,8,10,9,11,9,11,10,12,12,5,13,4,6,6,13,8,9,10,13,13,
11,10,5,3,3,4,9,6,8,3,5,3,2,2,1,3,5,11,2,3,5,6,9,8,5,
2,5,3,4,6,4,8,15,12,16,20,18,23,18,19,24,23,24,21,26,
36,38,37,39,45,42,41,37,38,38,35,37,35,31,32,30,20,39,
40,33,32,35,34,36,34,32,33,27,28,25,22,17,18,16,10,9,
5,12,7,8,8,9,19,21,24,20,23,19,17,18,17,22,11,12,3,9,
10,4,5,13,3,5,6,3,5,4,2,5,1,2,4,4,3,2,1)
###############################################################
## Use a (m=4 state) hidden Markov model based on the #########
## generalized poisson distribution to assign an #########
## activity range to the counts #########
###############################################################
## In this example three activity ranges ####
## (named as "light", "moderate" and "vigorous" physical ####
## activity) are separated by the two cut-points 15 and 23 ####
###############################################################
HMM_based_cut_off_point_method <- HMM_based_method(x = x,
cut_points = c(15,23),
min_m = 4,
max_m = 4,
names_activity_ranges = c("LIG","MOD","VIG"),
distribution_class = "genpois",
training_method = "numerical",
DNM_limit_accuracy = 0.05,
DNM_max_iter = 10,
bout_lengths = c(1,1,2,4,5,10,11,20,21,60,61,260),
plotting = 1)
###############################################################
## Example 4) In a single function (Step 1-3 of Example 2 ####
## combined in one function) ####
## (large and highly scatterd time-series) ####
###############################################################
################################################################
### Generate a large time-series of highly scattered counts ####
################################################################
x <- HMM_simulation(
size = 1500,
m = 10,
gamma = 0.93 * diag(10) + rep(0.07 / 10, times = 10),
distribution_class = "norm",
distribution_theta = list(mean = c(10, 100, 200, 300, 450,
600, 700, 900, 1100, 1300, 1500),
sd=c(rep(100,times=10))),
obs_round=TRUE,
obs_non_neg=TRUE,
plotting=5)$observations
################################################################
### Compare results of the tradional cut-point method ##########
### and the (6-state-normal-)HMM based method ##################
################################################################
traditional_cut_off_point_method <- cut_off_point_method(x=x,
cut_points = c(200,500,1000),
names_activity_ranges = c("SED","LIG","MOD","VIG"),
bout_lengths = c(1,1,2,4,5,10,11,20,21,60,61,260),
plotting = 1)
HMM_based_cut_off_point_method <- HMM_based_method(x=x,
max_scaled_x = 200,
cut_points = c(200,500,1000),
min_m = 6,
max_m = 6,
BW_limit_accuracy = 0.5,
BW_max_iter = 10,
names_activity_ranges = c("SED","LIG","MOD","VIG"),
distribution_class = "norm",
bout_lengths = c(1,1,2,4,5,10,11,20,21,60,61,260),
plotting = 1)
HMMpa documentation built on May 2, 2019, 7:58 a.m.
R Package Documentation
Browse R Packages
We want your feedback!
Note that we can't provide technical support on individual packages. You should contact the package authors for that.
Description Details Note Author(s) References Examples
Description
This package provides functions for analyzing accelerometer outpout data (known as a time-series of (impulse)-counts) to quantify length and intensity of physical activity.
Usually, so called activity ranges are used to classify an activity as “sedentary”, “moderate” and so on. Activity ranges are separated by certain thresholds (cut-off points). The choice of these cut-off points depends on different components like the subjects' age or the type of accelerometer device.
Cut-off point values and defined activity ranges are important input values of the following analyzing tools provided by this package:
1. Cut-off point method (assigns an activity range to a count given its total magintude). This traditional approach assigns an activity range to each count of the time-series independently of each other given its total magnitude.
2. HMM-based method (assigns an activity range to a count given its underlying PA-level). This approach uses a stochastic model (the hidden Markov model or HMM) to identify the (Markov dependent) time-series of physical activity states underlying the given time-series of accelerometer counts. In contrast to the cut-off point method, this approach assigns activity ranges to the estimated PA-levels corresponding to the hidden activity states, and not directly to the accelerometer counts.
Details
| Package: | HMMpa |
| Type: | Package |
| Version: | 1.0.1 |
| Date: | 2018-04-20 |
| License: | GPL-3 |
The new procedure for analyzing accelerometer data can be roughly described as follows:
First, a hidden Markov model (HMM) is trained to estimate the number m of hidden physical activity states and the model specific parameters (delta, gamma, distribution_theta). Then, a user-sepcified decoding algorithm decodes the trainded HMM to classify each accelerometer count into the m hidden physical activity states. Finally, the estimated distribution mean values (PA-levels) corresponding to the hidden physical activity states are extracted and the accelerometer counts are assigned by the total magnitudes of their corresponding PA-levels to given physical activity ranges (e.g. "sedentary", "light", "moderate" and "vigorous") by the traditional cut-off point method.
Note
We thank Moritz Hanke for his help in realizing this package.
Author(s)
Vitali Witowski,
Ronja Foraita,
Leibniz Institute for Prevention Research and Epidemiology (BIPS)
Maintainer: Ronja Foraita <foraita@leibniz-bips.de>
References
Baum, L., Petrie, T., Soules, G., Weiss, N. (1970). A maximization technique occurring in the statistical analysis of probabilistic functions of markov chains. The annals of mathematical statistics, vol. 41(1), 164–171.
Brachmann, B. (2011). Hidden-Markov-Modelle fuer Akzelerometerdaten. Diploma Thesis, University Bremen - Bremen Institute for Prevention Research and Social Medicine (BIPS).
Dempster, A., Laird, N., Rubin, D. (1977). Maximum likelihood from incomplete data via the EM algorithm. Journal of the Royal Statistical Society. Series B (Methodological), vol. 39(1), 1–38.
Forney, G.D. (1973). The Viterbi algorithm. Proceeding of the IEE, vol. 61(3), 268–278.
Joe, H., Zhu, R. (2005). Generalized poisson distribution: the property of mixture of poisson and comparison with negative binomial distribution. Biometrical Journal, vol. 47(2), 219–229.
MacDonald, I. L., Zucchini, W. (2009) Hidden Markov Models for Time Series: An Introduction Using R, Boca Raton: Chapman & Hall.
Viterbi, A.J. (1967). Error Bounds for concolutional codes and an asymptotically optimal decoding algorithm. Information Theory, IEEE Transactions on, vol. 13(2), 260–269.
Witowski, V., Foraita, R., Pitsiladis, Y., Pigeot, I., Wirsik, N. (2014) Using hidden Markov models to improve quantifying physical activity in accelerometer data - A simulation study. PLOS ONE. 9(12), e114089. http://dx.doi.org/10.1371/journal.pone.0114089
Examples
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162 163 164 165 166 167 168 169 170 171 172 173 174 175 176 177 178 179 180 181 182 183 184 185 186 187 188 189 190 191 192 193 194 195 196 197 198 199 200 201 202 203 | ################################################################
### Example 1 (traditional approach) ###################
### Solution of the cut-off point method ###################
################################################################
################################################################
### Fictitious activity counts #################################
################################################################
x <- c(1,16,19,34,22,6,3,5,6,3,4,1,4,3,5,7,9,8,11,11,
14,16,13,11,11,10,12,19,23,25,24,23,20,21,22,22,18,7,
5,3,4,3,2,3,4,5,4,2,1,3,4,5,4,5,3,5,6,4,3,6,4,8,9,12,
9,14,17,15,25,23,25,35,29,36,34,36,29,41,42,39,40,43,
37,36,20,20,21,22,23,26,27,28,25,28,24,21,25,21,20,21,
11,18,19,20,21,13,19,18,20,7,18,8,15,17,16,13,10,4,9,
7,8,10,9,11,9,11,10,12,12,5,13,4,6,6,13,8,9,10,13,13,
11,10,5,3,3,4,9,6,8,3,5,3,2,2,1,3,5,11,2,3,5,6,9,8,5,
2,5,3,4,6,4,8,15,12,16,20,18,23,18,19,24,23,24,21,26,
36,38,37,39,45,42,41,37,38,38,35,37,35,31,32,30,20,39,
40,33,32,35,34,36,34,32,33,27,28,25,22,17,18,16,10,9,
5,12,7,8,8,9,19,21,24,20,23,19,17,18,17,22,11,12,3,9,
10,4,5,13,3,5,6,3,5,4,2,5,1,2,4,4,3,2,1)
traditional_cut_off_point_method <- cut_off_point_method(x=x,
cut_points = c(5,15,23),
names_activity_ranges = c("SED","LIG","MOD","VIG"),
bout_lengths = c(1,1,2,4,5,10,11,20,21,60,61,260),
plotting = 1)
################################################################
################################################################
### Examples 2,3 and 4 (new approach) ########
### Solution of the HMM based cut-off point method ########
################################################################
### Demonstrated both in three steps (Example 2) ######
### and condensed in one function (Examples 3 and 4) ######
################################################################
################################################################
### Example 2) Manually in three steps ######################
################################################################
################################################################
### Fictitious activity counts #################################
################################################################
x <- c(1,16,19,34,22,6,3,5,6,3,4,1,4,3,5,7,9,8,11,11,
14,16,13,11,11,10,12,19,23,25,24,23,20,21,22,22,18,7,
5,3,4,3,2,3,4,5,4,2,1,3,4,5,4,5,3,5,6,4,3,6,4,8,9,12,
9,14,17,15,25,23,25,35,29,36,34,36,29,41,42,39,40,43,
37,36,20,20,21,22,23,26,27,28,25,28,24,21,25,21,20,21,
11,18,19,20,21,13,19,18,20,7,18,8,15,17,16,13,10,4,9,
7,8,10,9,11,9,11,10,12,12,5,13,4,6,6,13,8,9,10,13,13,
11,10,5,3,3,4,9,6,8,3,5,3,2,2,1,3,5,11,2,3,5,6,9,8,5,
2,5,3,4,6,4,8,15,12,16,20,18,23,18,19,24,23,24,21,26,
36,38,37,39,45,42,41,37,38,38,35,37,35,31,32,30,20,39,
40,33,32,35,34,36,34,32,33,27,28,25,22,17,18,16,10,9,
5,12,7,8,8,9,19,21,24,20,23,19,17,18,17,22,11,12,3,9,
10,4,5,13,3,5,6,3,5,4,2,5,1,2,4,4,3,2,1)
################################################################
## Step 1: Training of a HMM ##################################
## for the given time-series of counts ################
## ####################################################
## More precisely: training of a poisson ##############
## distribution based hidden Markov model for #########
## number of states m=2,...,6 #########################
## and selection of the model with the most ###########
## plausible m ########################################
################################################################
m_trained_HMM <- HMM_training(x = x,
min_m = 2,
max_m = 6,
distribution_class = "pois")$trained_HMM_with_selected_m
###############################################################
## Step 2: Decoding of the trained HMM for the given ##########
## time-series of accelerometer counts to extract #####
## hidden PA-levels ###################################
###############################################################
hidden_PA_levels <- HMM_decoding(x = x,
m = m_trained_HMM$m,
delta = m_trained_HMM$delta,
gamma = m_trained_HMM$gamma,
distribution_class = m_trained_HMM
$distribution_class,
distribution_theta = m_trained_HMM$
distribution_theta)$decoding_distr_means
###############################################################
## Step 3: Assigning of user-sepcified activity ranges ########
## to the accelerometer counts via the total ##########
## magnitudes of their corresponding ##################
## hidden PA-levels ###################################
## ####################################################
## In this example four activity ranges ###############
## (named as "sedentary", "light", "moderate" #########
## and "vigorous" physical activity) are ##############
## separated by the three cut-points 5, 15 and 23) ####
################################################################
HMM_based_cut_off_point_method <- cut_off_point_method(x = x,
hidden_PA_levels = hidden_PA_levels,
cut_points = c(5,15,23),
names_activity_ranges = c("SED","LIG","MOD","VIG"),
bout_lengths = c(1,1,2,4,5,10,11,20,21,60,61,260),
plotting=1)
###############################################################
## Example 3) In a single function (Step 1-3 of Example 2 ####
## combined in one function) ####
###############################################################
###############################################################
## Fictitious activity counts #################################
###############################################################
x <- c(1,16,19,34,22,6,3,5,6,3,4,1,4,3,5,7,9,8,11,11,
14,16,13,11,11,10,12,19,23,25,24,23,20,21,22,22,18,7,
5,3,4,3,2,3,4,5,4,2,1,3,4,5,4,5,3,5,6,4,3,6,4,8,9,12,
9,14,17,15,25,23,25,35,29,36,34,36,29,41,42,39,40,43,
37,36,20,20,21,22,23,26,27,28,25,28,24,21,25,21,20,21,
11,18,19,20,21,13,19,18,20,7,18,8,15,17,16,13,10,4,9,
7,8,10,9,11,9,11,10,12,12,5,13,4,6,6,13,8,9,10,13,13,
11,10,5,3,3,4,9,6,8,3,5,3,2,2,1,3,5,11,2,3,5,6,9,8,5,
2,5,3,4,6,4,8,15,12,16,20,18,23,18,19,24,23,24,21,26,
36,38,37,39,45,42,41,37,38,38,35,37,35,31,32,30,20,39,
40,33,32,35,34,36,34,32,33,27,28,25,22,17,18,16,10,9,
5,12,7,8,8,9,19,21,24,20,23,19,17,18,17,22,11,12,3,9,
10,4,5,13,3,5,6,3,5,4,2,5,1,2,4,4,3,2,1)
###############################################################
## Use a (m=4 state) hidden Markov model based on the #########
## generalized poisson distribution to assign an #########
## activity range to the counts #########
###############################################################
## In this example three activity ranges ####
## (named as "light", "moderate" and "vigorous" physical ####
## activity) are separated by the two cut-points 15 and 23 ####
###############################################################
HMM_based_cut_off_point_method <- HMM_based_method(x = x,
cut_points = c(15,23),
min_m = 4,
max_m = 4,
names_activity_ranges = c("LIG","MOD","VIG"),
distribution_class = "genpois",
training_method = "numerical",
DNM_limit_accuracy = 0.05,
DNM_max_iter = 10,
bout_lengths = c(1,1,2,4,5,10,11,20,21,60,61,260),
plotting = 1)
###############################################################
## Example 4) In a single function (Step 1-3 of Example 2 ####
## combined in one function) ####
## (large and highly scatterd time-series) ####
###############################################################
################################################################
### Generate a large time-series of highly scattered counts ####
################################################################
x <- HMM_simulation(
size = 1500,
m = 10,
gamma = 0.93 * diag(10) + rep(0.07 / 10, times = 10),
distribution_class = "norm",
distribution_theta = list(mean = c(10, 100, 200, 300, 450,
600, 700, 900, 1100, 1300, 1500),
sd=c(rep(100,times=10))),
obs_round=TRUE,
obs_non_neg=TRUE,
plotting=5)$observations
################################################################
### Compare results of the tradional cut-point method ##########
### and the (6-state-normal-)HMM based method ##################
################################################################
traditional_cut_off_point_method <- cut_off_point_method(x=x,
cut_points = c(200,500,1000),
names_activity_ranges = c("SED","LIG","MOD","VIG"),
bout_lengths = c(1,1,2,4,5,10,11,20,21,60,61,260),
plotting = 1)
HMM_based_cut_off_point_method <- HMM_based_method(x=x,
max_scaled_x = 200,
cut_points = c(200,500,1000),
min_m = 6,
max_m = 6,
BW_limit_accuracy = 0.5,
BW_max_iter = 10,
names_activity_ranges = c("SED","LIG","MOD","VIG"),
distribution_class = "norm",
bout_lengths = c(1,1,2,4,5,10,11,20,21,60,61,260),
plotting = 1)
|
R Package Documentation
Browse R Packages
We want your feedback!
Note that we can't provide technical support on individual packages. You should contact the package authors for that.
Embedding an R snippet on your website
Add the following code to your website.
For more information on customizing the embed code, read Embedding Snippets.