clusteringTypicalDays: Generate a set of flow-based typical days
In rte-antares-rpackage/flowBasedClustering: Cluster 'Flow-Based' Typical Days

Run a clustering algorithm on the different classes of the calendar (getCalendar). Its principle is to create clusters by gathering the most similar days of each class and to choose among them the best representative: it will be a so-called typical day. The metric used to determine the similarity of two days is a weighted sum of 24 hourly distances, meaning the distances between the domains of the two days at the same hour.

1
2
3

clusteringTypicalDays(calendar, vertices, nbClustWeek = 3,
  nbClustWeekend = 1, report = TRUE, reportPath = getwd(),
  hourWeight = rep(1, 24), maxDomainSize = 10000)

`calendar`	`list`, vector of date for each period. Can be obtain with getCalendar
`vertices`	`data.table`, 5 columns : Date : date ( Period : Hour (1:24), period i corresponds to the hourly time-step [i-1 ; i] BE : belgium vertices DE : german vertices FR : french vertices This parameter can be obtained with the function ptdfToVertices.
`nbClustWeek`	`numeric`, number of clusters for week period(working days). Defaut to 3
`nbClustWeekend`	`numeric`, number of clusters for weekend period. Defaut to 1
`report`	`boolean`, should a .html report be generated with the results of the clustering. Defaut to TRUE
`reportPath`	`character`, path where the report is written. Defaut to `getwd()`
`hourWeight`	`numeric`, vector of 24 weights, one for each hour of the day. The clustering algorithm will be more accurate for the flow-based domains of the hours with a relatively higher weight.
`maxDomainSize`	`numeric` limit of domain size in each axis. The function will return an error if one domain or more exceed these limits.

library(data.table)

# read vertices (from file, or obtained them with ptdfToVertices() function)
vertices <- fread(system.file("dataset/vertices_example.txt",package = "flowBasedClustering"))

# build calendar (manually here to adapt to the small dataset in example)
 # (for pratical applications getCalendar() function might be more convenient)
 
calendar <- list()
calendar$interSeasonWe <- c("2016-09-17", "2016-09-18")
calendar$interSeasonWd <- c("2016-09-19", "2016-09-20", "2016-09-21", "2016-09-22", "2016-09-23")
calendar$winterWe <- c("2016-12-10", "2016-12-11")
calendar$winterWd <- c("2016-12-12", "2016-12-13", "2016-12-14", "2016-12-15", "2016-12-16")
calendar$summerWe <- c("2016-08-06", "2016-08-07")
calendar$summerWd <- c("2016-08-08", "2016-08-09", "2016-08-10", "2016-08-11", "2016-08-12")

# run clustering algorithm
clusterTD <- clusteringTypicalDays(calendar, vertices, nbClustWeek = 2, nbClustWeekend = 1)

# run clustering algorithm distinguishing the days only with the flow-based domains of hour 18
w <- rep(0,24)
w[18] <- 1
clusterTD <- clusteringTypicalDays(calendar, vertices,  hourWeight = w, report = FALSE)