# http://tsigeto.info/covid19/remcat20200705answer.r.txt
# For https://remcat.hatenadiary.jp/entry/20200705/answer
cluster.n <- 61
cluster <- data.frame(
	size = c( "5-10", "11-20", "21-30", "31-"),
	n    = c(  39,      16,       2,      4 ),
	mid  = c(   7.5,    15.5,    25.5,   35.5 )
)
cluster.case <- sum(cluster$mid * cluster$n)
cluster.size.av <- cluster.case / cluster.n
cluster.size.av


byweek <- data.frame(
	epiweek = 7:14,
	date = c( "-Feb15", "Feb16-22", "Feb23-29", "Mar1-7", "Mar8-14", "Mar15-21", "Mar22-28", "Mar29-Apr4") ,
	cluster.n = c(     3,  3,   4,   4,   7,   7,  15,   18 ),
	link.ratio= c( 0.80,0.94,0.66,0.83,0.90,0.75,0.70, 0.45 ),
	import.case=c( 12+ 5,  1,   2,   4,  20,  55, 113,   97 ),
	local.case= c(  3+23, 72, 105, 214, 305, 198, 517, 1438 ),
	all.case  = c( 15+28, 73, 107, 218, 325, 253, 630, 1535 ), 
	toyo.case = c( 16+30, 68, 110, 215, 323, 253, 631, 1398 ) 	# From ToyoKeizai data
)

byweek$all.case - byweek$toyo.case


byweek$cluster.case<- byweek$cluster.n * cluster.size.av
byweek$link.case   <- byweek$local.case* byweek$link.ratio
byweek$nolink.case <- byweek$local.case   - byweek$link.case
byweek$noncluster.case <- byweek$link.case- byweek$cluster.case

case.byweek <- data.frame (
	epiweek = byweek$epiweek,
	date = byweek$date,
	cluster = byweek$cluster.case,
	noncluster = byweek$noncluster.case,
	nolink = byweek$nolink.case
)

case.byweek$cluster[1] <- 20.8
case.byweek$noncluster[1] <- 0

case.byweek$total <- case.byweek$cluster + case.byweek$noncluster + case.byweek$nolink
case.byweek


pct.byweek <- case.byweek
pct.byweek$cluster   <- round( 100 * case.byweek$cluster    / case.byweek$total , 1)
pct.byweek$noncluster<- round( 100 * case.byweek$noncluster / case.byweek$total , 1)
pct.byweek$nolink    <- round( 100 * case.byweek$nolink     / case.byweek$total , 1)
pct.byweek$total    <- round( 100 * case.byweek$total     / case.byweek$total , 1)
pct.byweek


cluster <- sum( case.byweek$cluster[c(1:3)] )
noncluster <- sum( case.byweek$noncluster[c(1:3)] )
cluster / (cluster + noncluster)

cluster <- sum( case.byweek$cluster[c(4:7)] )
noncluster <- sum( case.byweek$noncluster[c(4:7)] )
cluster / (cluster + noncluster)