OWL - Optimal Warehouse Locator
A brief introduction on how the problem is constructed, and the business problem targeted to solve.
library(skm)
library(plyr)
library(dplyr)
#- question: a busniess want to start building it's warehouses - objective reach
#- as many as u.s population as possible - minimize population weighted distance
#- so where to build the first one, the second one, the third one, and etc.?
## load skm built in datasets
data("zip2012")
data("source_zip_list")
## suppose build at most one warehouse in each state located at source_zip_list:
## create a s <source> - t <destination> - dist <distance> analytical data.table
## create dsrc s <source>
dsrc <- zip2012 %>%
subset(zip %in% source_zip_list) %>%
rename(s = zip, s_lat = lat, s_lng = lng) %>%
select(s, s_lat, s_lng) %>%
`class<-`(c("data.table", "data.frame"))
## create ddst t <target> (destination)
## want to reach all u.s. so all zip included
ddst <- zip2012 %>%
rename(t = zip, t_lat = lat, t_lng = lng) %>%
select(t, t_lat, t_lng, pop) %>%
`class<-`(c("data.table", "data.frame"))
## create analytical data ddzt: s - t - dist
# > yg::CJ.dt
# function (X, Y)
# {
# stopifnot(is.data.table(X), is.data.table(Y))
# k = NULL
# X = X[, c(k = 1, .SD)]
# setkey(X, k)
# Y = Y[, c(k = 1, .SD)]
# setkey(Y, NULL)
# return(X[Y, allow.cartesian = TRUE][, `:=`(k, NULL)][])
# }
# <environment: namespace:yg>
ddzt <- yg::CJ.dt(dsrc, ddst) %>%
mutate(dist = distRpl_wlatlng_cpp(s_lat, s_lng, t_lat, t_lng)) %>%
`class<-`(c("data.table", "data.frame"))
## solve the problem with skm_mls
## suppose we want to know the best solution for 1-10 number of warehouse:
skm_mls(
x = ddzt, k = 1L:10L, s_colname = "s", t_colname = "t",
d_colname = "dist", w_colname = "pop"
)
## suppose we already built 4 warehouses at NJ, IA, TX and CA, and want to
## know where to build the 5th - 10th warehouse:
skm_mls(
x = ddzt, k = 5L:10L, s_colname = "s", t_colname = "t",
d_colname = "dist", w_colname = "pop",
s_must = c("07017", "50317", "77804", "94596")
)
## a slightly different scenario is that we decided to build 4 warehouses
## at NJ, IA, TX and CA, and also want to know where to build the 5-10th.
## when number of warehouse <= 4, select location within 4 already built,
## and when number of warehhouse > 4 we build on top of these 4 locations
skm_mls(
x = ddzt, k = 1L:10L, s_colname = "s", t_colname = "t",
d_colname = "dist", w_colname = "pop",
s_must = c("07017", "50317", "77804", "94596")
)