mainscntanalysis/secondDbSteps.r

74 lines
2.0 KiB
R

library(DBI)
library(tidyr)
library(dplyr)
library(lubridate)
library(R.utils)
get_freq_df <- function(con, startDate, endDate) {
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)
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
}
}
return (freq_wide)
}
con <- dbConnect(RPostgres::Postgres(),
dbname='mainscnt',
host='db.mainscnt.eu',
user='wn')
START <- "2021-08-03 07:00:00"
INTERVAL <- 3600
for (offset in 0:0) {
startDate <- ymd_hms(START) + INTERVAL * offset
endDate <- startDate + INTERVAL
freq_wide <- get_freq_df(con, startDate, endDate)
freq_wide_names <- names(freq_wide)
for (colIdx in 2:length(freq_wide)) {
colName.mean <- paste("mean.w.o.", freq_wide_names[colIdx], sep="")
colName.diff <- paste(freq_wide_names[colIdx], ".to.mean", sep="")
freq_wide <- freq_wide %>%
rowwise() %>%
mutate(!!colName.mean := mean(c_across(freq_wide_names[c(-1, - colIdx)]), na.rm=TRUE)) %>%
mutate(!!colName.diff := abs(eval(as.name(colName.mean)) - eval(as.name(freq_wide_names[colIdx]))) * 100)
}
means <- freq_wide %>% select(ends_with(".to.mean"))
means.means <- apply(means, 2, mean, na.rm=TRUE)
# print(summary(freq_wide))
}
dbDisconnect(con)