test/classes.R

In maxkroeschel/activitytools: Tools to analyse and visualize activity data in R.

library(devtools)
load_all("./")
Sys.setenv(TZ='UTC')

# Test activity class ----
data(activity_data)
data(gps_data)

activity_data$act_xy <- activity_data$act_x + activity_data$act_y

pars <- list(act.act = "act_xy",
             act.reg_minutes = 5,
             act.width_ma = 2,
             tresh.n_runs = 1,
             thresh.window_width_around_day = 3,
             thresh.n_thresholds = c(25:35),
             thresh.resting_range_limit = 0.2,
             thresh.threshold_range_limit = 0.8,
             states.min_duration_active = 10)
# simple
deer <- activity(activity_data = activity_data)
# with GPS
deer <- activity(activity_data = activity_data,
                 gps = gps_data)
# with parameters
deer <- activity(activity_data = activity_data,
                 gps = gps_data,
                 parameters = pars)
# based on other activity object
deeragain <- activity(activity_data = deer$activity_data,
                      gps = deer$gps_data,
                      parameters = deer$parameters)
# with only some parameters
deer <- activity(activity_data = activity_data,
                 gps = gps_data,
                 parameters = pars[1:2])
# Don't keep source data
deer <- activity(activity_data = activity_data,
                 gps = gps_data,
                 parameters = pars,
                 keep_source = FALSE)
# Timestamp not POSIX
activity_data_noPOSIX <- activity_data
activity_data_noPOSIX$ts <- as.character(activity_data$ts)
deer <- activity(activity_data = activity_data_noPOSIX)
deer$activity_data$ts

# Throw some errors
deer <- activity(activity_data = NULL,
                 gps = gps_data)
deer <- activity(activity_data = activity_data[,c(1,3)])
deer <- activity(activity_data = activity_data,
                 gps = gps_data[,c(1,4)])


## Workflow 1 - all model parameters first ----
data(activity_data)
# reduce data set for fast testing
# activity_data <- activity_data[activity_data$ts >= as.POSIXct("2009-01-01") &
#                                  activity_data$ts <= as.POSIXct("2009-03-01"),]
data(gps_data)

# list of all parameters
pars <- list(act.act = "act_xy",
             act.reg_minutes = 5,
             act.width_ma = 2,
             thresh.n_runs = 1,
             thresh.window_width_around_day = 3,
             thresh.n_thresholds = c(25:35),
             thresh.resting_range_limit = 0.2,
             thresh.threshold_range_limit = 0.8,
             states.min_duration_active = 10,
             pta.pos = NULL,
             pta.dayshift = "dawn",
             pta.dawn_degree = 12,
             pta.period = "day",
             pta.max_na = 30)

deer <- activity(activity_data = activity_data, gps_data = gps_data, parameters = pars)

plot(deer, select = "activity_eval", select_animal_id = 1)
# Regularize
# deer$activity_data[,.(min = min(diff(ts)),
#                  max = max(diff(ts)),
#                  mean = mean(diff(ts))),
#               by = animal_tag]

deer <- regularize_activity(deer)

# deer$activity_data[,.(min = min(diff(ts)),
#                       max = max(diff(ts)),
#                       mean = mean(diff(ts))),
#                    by = animal_tag]

# Combine activity_data from x- and y-axis: Already performed when creating
# activity obect; updated after regularization
# deer[["active_states"]]

# Smooth activity
deer <- smooth_activity(deer)

# Identify and remove activity gaps
deer <- remove_activity_gaps(deer)

# Thresholds
deer <- calculate_thresholds(deer, plot_summary = TRUE)
# Load resulting 'deer' object:
#data("thresholds")

# Activity states (using parameter list defined above)
deer <- calculate_states(deer, add = c("pta", "gps"))

# States again with different parameters
deer2 <- states(activity = deer,
                thresholds = c("a", "c"),
                parameters = list(pta.dayshift = "sunrise",
                                  pta.dawn_degree = 11,
                                  pta.period = "week",
                                  pta.max_na = 210))

## Plotting
# Plot thresholds and activity
# Thresholds
plot(deer, select = "thresholds")
#Activity
plot(deer, select_animal_id = 1)
plot(deer)

plot(deer, select = "activity_eval", select_animal_id = 1)
plot(deer, select = "activity", select_animal_id = 1) # same
# Active states
plot(deer, select="states", threshold = "a")

# Export classified GPS data
write.table(deer_states$threshold_a$gps_active, file = "./test/gps_data_export.csv",
            sep  = ";", row.names = FALSE)


## Workflow 2 - parameters step by step ----
data(activity_data)
data(gps_data)
deer <- activity(activity_data = activity_data, gps = gps_data)

# Regularize
deer$activity_data[,.(min = min(diff(ts)),
                      max = max(diff(ts)),
                      mean = mean(diff(ts))),
                   by = animal_tag]

deer <- regularize_activity(activity = deer,
                            act.reg_minutes=5)
deer$activity_data[,.(min = min(diff(ts)),
                      max = max(diff(ts)),
                      mean = mean(diff(ts))),
                   by = animal_tag]

# Combine activity_data from x- and y-axis: Already performed when creating
# activity obect; updated after regularization

# Smooth activity
deer <- smooth_activity(activity = deer,
                        act.act = 'act_xy',
                        act.width_ma = 2,
                        update_NA = TRUE)

# Identify and remove activity gaps
deer <- identify_activity_gaps(deer,
                               act.act = 'act_xy')
deer <- remove_activity_gaps(deer)

# Thresholds -- includes aggregation
deer <- calculate_thresholds(activity = deer,
                              thresh.n_runs = 1,
                              thresh.window_width_around_day = 3,
                              thresh.n_thresholds = c(25:35),
                              thresh.resting_range_limit = 0.2,
                              thresh.threshold_range_limit = 0.8,
                              states.min_duration_active = 10,
                              plot_summary = TRUE)
data("thresholds")

# Activity states
deer_states <- calculate_states(deer,
                                add = c("pta", "gps")
                                thresholds = c("a", "b", "c"),
                                pta.dayshift = "dawn",
                                pta.dawn_degree = 12,
                                pta.period = "day",
                                pta.max_na = 30)

# using different parameters
deer_states2 <- calculate_states(activity = deer,
                                 add = c("pta", "gps")
                                 thresholds = c("a", "c"),
                                 pta.dayshift = "sunrise",
                                 pta.dawn_degree = 11,
                                 pta.period = "week",
                                 pta.max_na = 210)

# Plotting same as above
# Thresholds
plot(deer, select = "thresholds")
#Activity
plot(deer, animal_id = 1)
plot(deer, select = "activity", animal_id = 1) # same
# Active states
plot(deer, select="states", threshold = "b")

# GPS data ----

# GPS data can also be added or changed at later stages

deer <- add_gps_data(deer, gps_data)
deer_states <- add_gps_data(deer_states, gps_data)