02_preprocess_gasprice_data.Rmd

---
title: "Connecting to PostgreSQL"
author: "Henning"
output: html_document
---


# Dataprep
### Inspect and merge old data for Aral-Timeseries
```{r}
library(xts)
merge_ts <- function(ts_old, ts_akt, col = "e10"){
  ts_old <- xts(x = ts_old[col], order.by = ts_old$date)
  ts_akt <- xts(x = ts_akt[col], order.by = ts_akt$date)
  ts_merged <- rbind(ts_old, ts_akt)
  return(ts_merged)
}

diesel <- merge_ts(ts_old, ts_new, col = "diesel")
```


## Put the data on a hourly grid (take last price)
```{r}
on_hourly_grid <- function(ts){
# every 1 hours
s1 <- align.time(index(first(ts)), n = 3600)
s2 <- align.time(index(last(ts)),  n = 3600)
x1 <- seq(s1, s2, by = "1 hours")

d1 <- align.time(ts, n = 3600)

m2 <- cbind(d1, xts(order.by =  x1))
ts_hourlygrid <- na.locf(m2)
return(ts_hourlygrid)
}

diesel_h <- on_hourly_grid(diesel)
```

## Average over multiple prices on the same grid point
```{r}
avg_ts <- function(ts){
  ts_avg <- aggregate(ts, index(ts), function(d) mean(d))
  ts_avg <- as.xts(ts_avg)
  return(ts_avg)
}

diesel_h <- avg_ts(diesel_h)
saveRDS(diesel_h, "data/diesel_h.RDS")
```


# Aggregate Time Series
```{r}
# diesel_h
diesel_d <- apply.daily(diesel_h, maen, na.rm = TRUE)
diesel_w <- apply.weekly(diesel_h, mean, na.rm = TRUE)
diesel_m <- apply.monthly(diesel_h, mean, na.rm = TRUE)


d <- cbind(diesel_h, diesel_d, diesel_w, diesel_m)
plot(as.zoo(d))
```