From bee491a437b25f920e5a80c5dee8c99e7b39929f Mon Sep 17 00:00:00 2001
From: marjom02 <megan.mcnicol2@nhs.scot>
Date: Fri, 11 Aug 2023 13:21:52 +0100
Subject: [PATCH] Work in progress -

---
 R/process_lookup_homelessness.R | 119 ++++++++++++++++++++++++++++++++
 1 file changed, 119 insertions(+)
 create mode 100644 R/process_lookup_homelessness.R

diff --git a/R/process_lookup_homelessness.R b/R/process_lookup_homelessness.R
new file mode 100644
index 000000000..322dd798b
--- /dev/null
+++ b/R/process_lookup_homelessness.R
@@ -0,0 +1,119 @@
+# It would probably be simplest to create a lookup from the Homelessness data of one row per CHI +
+#   start dates of all applications. This can then be matched to episode and individual to easily create the hl1_in_fy flag(s)
+# (if someone is in the lookup -> 1, otherwise -> 0). For the episode file, it is slightly more complicated to create the other flags,
+# either matching on all the data by CHI and then doing some comparison of dates (as in SPSS),
+# or using some of the fancy new joins (dplyr 1.1.0 joins) and doing it that way.
+#
+#
+
+
+data <- slfhelper::read_slf_episode("1718", c("anon_chi", "record_keydate1", "record_keydate2", "recid"))
+
+year <-"1718"
+
+
+create_homelessness_lookup <- function(year) {
+  homelessness_lookup <- read_file(get_file_path(get_year_dir(year),
+                                                 stringr::str_glue("homelessness_for_source-20{year}"),
+                                                 ext = "rds",
+                                                 check_mode = "write")) %>%
+    dplyr::distinct(chi, record_keydate1, record_keydate2) %>%
+    tidyr::drop_na(chi) %>%
+    mutate(hl1_in_fy = 1) #%>%
+   # group_by(chi) %>%
+    #mutate(count = n())
+
+  return(homelessness_lookup)
+
+}
+
+
+add_homelessness_flag_episode <- function(data, year) {
+
+ lookup <- create_homelessness_lookup(year) %>%
+   slfhelper::get_anon_chi()
+
+ ## need to decide which recids this relates to
+ data1 <- data %>%
+   left_join(lookup %>%
+               distinct(anon_chi, hl1_in_fy),
+             by = "anon_chi", relationship = "many-to-one") %>%
+   mutate(hl1_in_fy = tidyr::replace_na(hl1_in_fy, 0))
+
+  return(data)
+
+}
+
+
+
+
+
+
+add_homelessness_date_flags_episode <- function(data, year) {
+
+
+  lookup <- create_homelessness_lookup(year) %>%
+    slfhelper::get_anon_chi() %>%
+    mutate(six_months_pre_app = record_keydate1 - lubridate::days(183),
+           six_months_post_app = record_keydate2 + lubridate::days(183))
+
+  data1 <- data %>%
+    select()
+   left_join(lookup %>%
+                distinct(anon_chi, hl1_in_fy, six_months_pre_app, six_months_post_app),
+              by = "anon_chi", relationship = "many-to-one") %>%
+    mutate(hl1_in_fy = tidyr::replace_na(hl1_in_fy, 0)) %>%
+ #   mutate(hl1_6after_ep = )
+ #
+ #
+ #
+ # hl1_during_ep    filter((application_date <= keydate2_dateformat) & (keydate1_dateformat <= end_date))
+ #
+ #
+ #
+ # Add hl1_6after_ep      filter((end_date <= keydate2_dateformat) &
+ #                                 (keydate1_dateformat <= six_months_post_app))
+ #
+ #
+ # Add hl1_6before_ep   ((six_months_pre_app <= keydate2_dateformat) &
+ #                         (keydate1_dateformat <= application_date))
+ #
+
+
+}
+
+
+
+# Numeric HH_ep  HH_6after_ep  HH_6before_ep (F1.0).
+# Variable Labels
+# HH_in_FY "CHI had an active homelessness application during this financial year"
+# HH_ep "CHI had an active homelessness application at time of episode"
+# HH_6after_ep "CHI had an active homelessness application at some point 6 months after the end of the episode"
+# HH_6before_ep "CHI had an active homelessness application at some point 6 months prior to the start of the episode".
+#
+# * I'm ignoring PIS (as the dates are not really episode dates), and CH as I'm not sure Care Homes tells us much (and the data is bad).
+# Do if any(recid, "00B", "01B", "GLS", "DD", "02B", "04B", "AE2", "OoH", "DN", "CMH", "NRS", "HL1").
+# Compute HH_ep = 0.
+# Compute HH_6after_ep = 0.
+# Compute HH_6before_ep = 0.
+#
+# * May need to change the numbers here depending on the max number of episodes someone has.
+# Do repeat AssessmentDecisionDate = AssessmentDecisionDate.1 to !maxAssessment.
+# * If there was an application decision made during episode.
+# * HH started during episode.
+# If Range(AssessmentDecisionDate, keydate1_dateformat, keydate2_dateformat) HH_ep = 1.
+#
+# * If there was an application decision made in the 6 months (180 days) after the episode discharged.
+# If Range(AssessmentDecisionDate, keydate2_dateformat + time.days(180), keydate2_dateformat + time.days(1)) HH_6after_ep = 1.
+#
+# * If the was an application decision made in the 6 months prior to admission.
+# If Range(AssessmentDecisionDate, keydate1_dateformat - time.days(180), keydate1_dateformat - time.days(1)) HH_6before_ep = 1.
+# End Repeat.
+# End if.
+#
+# If recid = 'HL1' and chi = '' HH_in_FY = 1 .
+#
+# *Save Temp.
+# save outfile = !Year_dir + "temp-source-episode-file-3-" + !FY + ".zsav"
+# /keep year to cij_delay HH_in_FY to HH_6before_ep
+# /zcompressed.