Photos
Blog
Projects
vanrenterghem.biz
projects
/
R
/
project-using-kafka-in-R.git
/ blobdiff
commit
grep
author
committer
pickaxe
?
search:
re
summary
|
shortlog
|
log
|
commit
|
commitdiff
|
tree
raw
|
inline
| side by side
Document SubscribeKafkaTopic function.
[R/project-using-kafka-in-R.git]
/
predictWebsiteHits.R
diff --git
a/predictWebsiteHits.R
b/predictWebsiteHits.R
index ec743109982a3dfa6e098fde2100603cf716238a..415ab1eebdedd0c32e88170e91b78bedd866f1b2 100644
(file)
--- a/
predictWebsiteHits.R
+++ b/
predictWebsiteHits.R
@@
-1,19
+1,24
@@
-setwd("~/work")
-library(jsonlite)
library(fable)
library(tsibble)
library(lubridate)
library(fable)
library(tsibble)
library(lubridate)
+library(ggplot2)
-logfile <- file("photos.vanrenterghem.biz.access.kvp.log")
-
-apachelog <- stream_in(logfile)
-
-#apachelog$time <- gsub("\\[|\\]", "", apachelog$time)
-apachelog %>%
+createPlot <- function(x){
+x %>%
mutate(time = gsub("\\[|\\]", "", time),
mutate(time = gsub("\\[|\\]", "", time),
- time = dmy_hms(time)) %>%
- group_by(time) %>%
+ time = dmy_hms(time),
+ datehour = floor_date(time, unit = "hour")
+ ) %>%
+ filter(time > ymd_h("2018-08-22 12")) %>%
+ group_by(datehour) %>%
summarise(hits = n()) ->
summarise(hits = n()) ->
- apachelog
+ apachelog
_tidy
-as.tsibble(apachelog, index = time)
+apachelog_tsbl <- as.tsibble(apachelog_tidy, index = datehour)
+apachelog_tsbl %>%
+ ARIMA(log(hits)) %>%
+ forecast(h=5) %>%
+ autoplot +
+ ggtitle("Forecasting website hits using apache log only") +
+ xlab("Date time")
+}
\ No newline at end of file