fixed *td functions for dbi

R/time_intelligence.R

@@ -372,6 +372,7 @@ totalytd <- function(.data,...,date_var,value_var){
 
   out_tbl <- full_tbl |>
     dplyr::group_by(year,...) |>
+    dplyr::arrange(date,.by_group = TRUE) |>
     dplyr::mutate(
     ytd=base::cumsum({{value_var}})
   ) |>
@@ -411,7 +412,8 @@ totalytd_dbi <- function(.data,...,date_var,value_var){
 
 
   out_dbi <- full_dbi |>
-    dbplyr::window_order(year,...) |>
+    dplyr::group_by(...,year) |>
+    dbplyr::window_order(date) |>
     dplyr::mutate(
       ytd=base::cumsum({{value_var}})
     ) |>
@@ -492,7 +494,8 @@ totalqtd_dbi <- function(.data,...,date_var,value_var){
 
 
   out_dbi <- full_dbi |>
-    dbplyr::window_order(year,quarter,...) |>
+    dplyr::group_by(year,quarter,...) |>
+    dbplyr::window_order(date) |>
     dplyr::mutate(
       qtd=base::cumsum({{value_var}})
     ) |>
@@ -576,11 +579,11 @@ totalmtd_dbi <- function(.data,...,date_var,value_var){
       ,month=lubridate::month(date)
     )
 
-  print("full_dbi completed")
 
 
   out_dbi <- full_dbi |>
-    dbplyr::window_order(year,month,...) |>
+    dplyr::group_by(year,month,...) |>
+    dbplyr::window_order(date) |>
     dplyr::mutate(
       mtd=base::cumsum({{value_var}})
     ) |>
@@ -666,7 +669,8 @@ totalwtd_dbi <- function(.data,...,date_var,value_var){
 
 
   out_dbi <- full_dbi |>
-    dbplyr::window_order(year,month,week,...) |>
+    dplyr::group_by(year,month,week,...) |>
+    dbplyr::window_order(date) |>
     dplyr::mutate(
       wtd=base::cumsum({{value_var}})
     ) |>
@@ -742,7 +746,8 @@ totalatd_dbi <- function(.data,...,date_var,value_var){
 
 
   out_dbi <- full_dbi |>
-    dbplyr::window_order(...) |>
+    dplyr::group_by(...) |>
+    dbplyr::window_order(date) |>
     dplyr::mutate(
       atd=base::cumsum({{value_var}})
     ) |>

README.qmd

@@ -65,7 +65,10 @@ format: gfm
 -   solved  logic for one group var vs. multiple group var
 -   created dbi equivalents all w,m,y,a todate functions
 -   remaining functions are wow,mom, and yoy
-
+
+2024-09-15
+- cleaned up window_order / group_by for dbi *td expressions
+
 ## to do list
 
 -   replicate time intelligence functions

glue.R

@@ -11,54 +11,79 @@ sales_db <- db$sales
 
 con <- dbplyr::remote_con(sales_db)
 
-.data <- sales_db
+.data <- sales_db |>
+  filter(
+    year(order_date)<=2016
+  )
+
+dbi_tbl <- make_aggregation_dbi(.data,customer_key,date_var=order_date,value_var =quantity,time_unit="day") |> pluck("dbi")
+df_tbl <- make_aggregation_tbl(sales |> filter(year(order_date)<=2016),customer_key,date_var=order_date,value_var =quantity,time_unit="day")
 
-totalatd_dbi(sales_db,date_var=order_date,value_var=quantity)
 
-totalatd(sales,date_var=order_date,value_var=quantity)
+totalytd_dbi(.data,customer_key,date_var=order_date,value_var=quantity) |>
+  filter(
+  customer_key=="210071"
+  # customer_key=="1979682"
+) |> view()
+  group_by(customer_key,year) |>
+  summarise(
+    max=max(ytd)
+    ,sum=sum(ytd)
+  )
 
-full_dbi <- sales_db |>
-  filter(year(order_date) %in% c("2015","2016")) |>
-  make_aggregation_dbi(customer_key,date_var=order_date,value_var=quantity,time_unit="day") |>
-  pluck("dbi")
 
+full_dbi <- dbi_tbl |>
+dplyr::mutate(
+  year=lubridate::year(date)
+  ,.before = 1
+)
 
-value_var <- "quantity"
 
 full_dbi |>
+  dbplyr::window_order(customer_key,date) |>
+  arrange(customer_key,date) |>
   dplyr::mutate(
-    date_lag=dplyr::sql("date + INTERVAL 1 WEEK")
-    ,"{{noquote(value_var)}}_wow":=quantity
-  )
-  dplyr::select(-c(date,{{value_var}})) |>
-  dplyr::ungroup()
-
-
-out_tbl <-  dplyr::left_join(
-  full_tbl
-  ,lag_table
-  ,by=dplyr::join_by(date==date_lag,...)
-) |>
-  mutate(
-    "{{value_var}}_wow" := dplyr::coalesce(.data[[rlang::englue("{{value_var}}_wow")]],0)
-  )
+    ytd=base::cumsum(quantity)
   ) |>
+
   dplyr::ungroup()
 
 
+all.equal(dbi_tbl,df_tbl)
 
-cout |>
-  pluck("dbi") |>
-  dplyr::mutate(
-    year=lubridate::year(date)
-    ,month=lubridate::quarter(date)
+
+full_join(
+  dbi_tbl
+  ,df_tbl
+  ,by=join_by(date,customer_key)
+) |>
+  group_by(
+    customer_key
   ) |>
-  dbplyr::window_order(year,month) |>
-  dplyr::mutate(
-    mtd=base::cumsum(quantity)
+  summarise(
+    dbi=sum(quantity.x)
+    ,df=sum(quantity.y)
+    ,n=n()
+    ,.groups = "drop"
   ) |>
-  dplyr::ungroup()
-
-
+  mutate(
+    delta=dbi-df
+  ) |>
+  summarise(
+    sum_delta=sum(delta)
+  )
+  # filter(customer_key=="2")
+
+  count(customer_key,wt="quantity.x")
+totalytd_dbi(.data,customer_key,date_var=order_date,value_var=quantity)
+totalatd_dbi(.data,customer_key,date_var=order_date,value_var=quantity)
+totalmtd_dbi(.data,customer_key,date_var=order_date,value_var=quantity)
+totalqtd_dbi(.data,customer_key,date_var=order_date,value_var=quantity) |>
+  group_by(customer_key,quarter,year) |>
+  summarise(
+    qtd=max(qtd)
+    ,sum=sum(quantity)
+  ) |> view()
+totalwtd_dbi(.data,customer_key,date_var=order_date,value_var=quantity)