Add resampling for fused location

config.yaml

@@ -46,6 +46,11 @@ PHONE_VALID_SENSED_DAYS:
   MIN_VALID_HOURS: 20 # (out of 24)
   MIN_BINS_PER_HOUR: 8 # (out of 60min/BIN_SIZE bins)
 
+RESAMPLE_FUSED_LOCATION:
+  CONSECUTIVE_THRESHOLD: 30 # minutes, only replicate location samples to the next sensed bin if the phone did not stop collecting data for more than this threshold
+  TIME_SINCE_VALID_LOCATION: 12 # hours, only replicate location samples to consecutive sensed bins if they were logged within this threshold after a valid location row
+  TIMEZONE: *timezone
+
 BARNETT_LOCATION:
   ACCURACY_LIMIT: 51 # filters location coordinates with an accuracy higher than this
   TIMEZONE: *timezone

rules/preprocessing.snakefile

@@ -51,4 +51,18 @@ rule unify_ios_android:
     output:
         "data/raw/{pid}/{sensor}_with_datetime_unified.csv"
     script:
-        "../src/data/unify_ios_android.R"
+        "../src/data/unify_ios_android.R"
+
+rule resample_fused_location:
+    input:
+        locations = "data/raw/{pid}/locations_raw.csv",
+        phone_sensed_bins = rules.phone_sensed_bins.output
+    params:
+        bin_size = config["PHONE_VALID_SENSED_DAYS"]["BIN_SIZE"],
+        timezone = config["RESAMPLE_FUSED_LOCATION"]["TIMEZONE"],
+        consecutive_threshold = config["RESAMPLE_FUSED_LOCATION"]["CONSECUTIVE_THRESHOLD"],
+        time_since_valid_location = config["RESAMPLE_FUSED_LOCATION"]["TIME_SINCE_VALID_LOCATION"]
+    output:
+        "data/raw/{pid}/locations_resampled.csv"
+    script:
+        "../src/data/resample_fused_location.R"

src/data/resample_fused_location.r

@@ -0,0 +1,43 @@
+source("packrat/init.R")
+
+
+library(dplyr)
+library(readr)
+library(tidyr)
+
+bin_size <- snakemake@params[["bin_size"]]
+timezone <- snakemake@params[["timezone"]]
+consecutive_threshold <- snakemake@params[["consecutive_threshold"]]
+time_since_valid_location <- snakemake@params[["time_since_valid_location"]]
+
+locations <- read_csv(snakemake@input[["locations"]], col_types = cols())
+phone_sensed_bins  <- read_csv(snakemake@input[["phone_sensed_bins"]], col_types = cols(local_date = col_character()))
+
+if(nrow(locations) > 0){
+    sensed_minute_bins <- phone_sensed_bins %>% 
+    pivot_longer(-local_date, names_to = c("hour", "bin"), names_ptypes = list(hour = integer(), bin = integer()), names_sep = "_", values_to = "sensor_count") %>% 
+    complete(nesting(local_date, hour), bin = seq(0, 59,1)) %>% 
+    fill(sensor_count) %>% 
+    mutate(timestamp = as.numeric(as.POSIXct(paste0(local_date, " ", hour,":", bin,":00"), format = "%Y-%m-%d %H:%M:%S", tz = timezone)) * 1000 ) %>%
+    filter(sensor_count > 0) %>% 
+    select(timestamp)
+
+    resampled_locations <- locations %>%
+    filter(provider == "fused") %>% 
+    bind_rows(sensed_minute_bins) %>% 
+    arrange(timestamp) %>% 
+    # We group and therefore, fill in, missing rows that appear after a valid fused location record and exist
+    # within consecutive_threshold minutes from each other
+    mutate(consecutive_time_diff = c(1, diff(timestamp)),
+         resample_group = cumsum(!is.na(double_longitude) | consecutive_time_diff > (1000 * 60 * consecutive_threshold))) %>% 
+    group_by(resample_group) %>% 
+    # drop rows that are logged after time_since_valid_location hours from the last valid fused location
+    filter((timestamp - first(timestamp) < (1000 * 60 * 60 * time_since_valid_location))) %>% 
+    fill(-timestamp, -resample_group) %>% 
+    select(-consecutive_time_diff) %>% 
+    drop_na(double_longitude, double_latitude, accuracy)
+
+    write.csv(resampled_locations,snakemake@output[[1]], row.names = F)
+} else {
+    write.csv(locations,snakemake@output[[1]], row.names = F)
+}