mainscntanalysis/secondDbSteps.r

library(DBI)
library(tidyr)
library(dplyr)
library(lubridate)
library(R.utils)

con <- dbConnect(RPostgres::Postgres(), 
                 dbname='mainscnt', 
                 host='db.mainscnt.eu', 
                 user='wn')


START <- "2021-08-03 00:00:00"
INTERVAL <- 3600

for (offset in 0:23) {
  startDate <- ymd_hms(START) + INTERVAL * offset
  endDate <- startDate + INTERVAL

  startStr <- strftime(startDate, "%Y-%m-%d %H:%M:%S", tz="UTC")
  endStr <- strftime(endDate, "%Y-%m-%d %H:%M:%S", tz="UTC")
  
  res <-dbSendQuery(con, "select time, location, freq from mainsfrequency where valid=1 and time >= $1 and time < $2")
  dbBind(res, list(startStr, endStr))
  frequencies <- dbFetch(res)
  dbClearResult(res)
  
  freq_wide <- frequencies %>% 
    pivot_wider(names_from = location, 
                values_from = freq, 
                values_fn = mean)
  
  freq_wide$Linz_AT <- freq_wide$Linz_AT + 0.002
  
  THRESHOLD <- 0.5
  
  for (colIdx in 2:length(freq_wide)) {
    last <- freq_wide[[1, colIdx]]
    for (rowIdx in 1:length(freq_wide[[colIdx]])) {
      current <- freq_wide[[rowIdx, colIdx]]
      if (!is.na(current) && !is.na(last) && (abs(current - last) > THRESHOLD)) {
        freq_wide[[rowIdx, colIdx]] = NA
      }
      last <- current
    }
  }
  
  
  freq_wide <- freq_wide %>% 
    rowwise() %>% 
    mutate(mad = mad(c_across(names(freq_wide)[-1]), 
                     na.rm=TRUE))
  
  # print(summary(freq_wide))
  printf("start: %s, end: %s, mad/cnt: %f\n", startStr, endStr, sum(freq_wide$mad) / length(freq_wide$mad))
}
  
dbDisconnect(con)