remove multiple assigne for one pattent in year/geoid

ruralinnovation · Sep 3, 2024 · 9861ee7 · 9861ee7
1 parent 540a7da
commit 9861ee7
Show file tree

Hide file tree

Showing 3 changed files with 24 additions and 6 deletions.
diff --git a/R/counties.R b/R/counties.R
@@ -199,12 +199,26 @@ get_patent_counts_wide <- function(patent_raw, cpc, assignee, location, cpc_code
                        by = "location_id")
   tidy_patents <- merge(patent_cpc_get2014, assignee_us, by = "patent_id")
   tidy_cori_patents <- merge(tidy_patents, cpc_codes, by = "cpc_subsection")
+  # if not adding assignee id it is not anymore a PK hence we will have dup
   slim_tidy_cori_patents <- tidy_cori_patents[, list(patent_id, year,
-                                                     cpc_subsection, assignee_id,
+                                                     cpc_subsection,
                                                      geoid_co)]
-  patent_counts_wide <- slim_tidy_cori_patents |>
+
+  slim_tidy_cori_patents_oli <- slim_tidy_cori_patents[!duplicated(slim_tidy_cori_patents),]
+
+  patent_counts_wide <- slim_tidy_cori_patents_oli |>
               as.data.frame() |>
               dplyr::mutate(value = 1) |>
               tidyr::pivot_wider(names_from = cpc_subsection,
                                  values_from = value)
+}
+
+write_to_proj_erc <- function(table_name, data, schema = "proj_erc") {
+  con <- cori.db::connect_to_db(schema)
+	message(schema)
+	on.exit(DBI::dbDisconnect(con))
+  (DBI::dbWriteTable(conn = con,
+                     name = table_name,
+                     value = data,
+                     overwrite = TRUE))
 }
diff --git a/_targets.R b/_targets.R
@@ -41,7 +41,9 @@ list(
   tar_target(location_us, get_me_us_location(location_raw)),
   tar_target(patent_counts_wide, get_patent_counts_wide(
     patent_raw, cpc, assignee_raw, location_us, cpc_codes_mem
-  ))
+  )),
+  tar_target(patent_db, write_to_proj_erc("patent_counts_wide",
+                                          patent_counts_wide))
   # tar_target(patent_assignee_location,
   #            get_me_patent_assignee_loc(patent_raw,
   #                                       assignee_raw,

diff --git a/_targets/meta/meta b/_targets/meta/meta
@@ -2,7 +2,7 @@ name|type|data|command|depend|seed|path|time|size|bytes|format|repository|iterat
 assignee_file|stem|62fc812b42f8aa00|ff2fad4cee4a8e2c|2c530c1562a7fbd1|-130791275|data/data_raw/unzipped/g_assignee_disambiguated.tsv|t19941.7783078089s|19181e6d5ae6bb8c|1037584231|file|local|vector|||0||
 assignee_raw|stem|17eea75cde4ed1c1|40f15c41a175cdbd|79a369d88be37457|-1776892158||t19965.6105602966s|408b0c8a8799d884|280998712|rds|local|vector|||8.411||
 cpc|stem|f473d9312543c832|69b2b2bd98e57276|dcb7b40df023f094|1232020923||t19968.7271009231s|4d909c8092cb94ec|71101564|rds|local|vector|||28.815||
-cpc_codes|stem|46ac3b5adf4523d4|aa8141e132cf2a5c|2c530c1562a7fbd1|-2144215752|data/CPC_patent_codes.csv|t19968.61136586s|8282224de3c290c3|1722|file|local|vector|||0.295||
+cpc_codes|stem|46ac3b5adf4523d4|aa8141e132cf2a5c|2c530c1562a7fbd1|-2144215752|data/CPC_patent_codes.csv|t19969.702277983s|8282224de3c290c3|1722|file|local|vector|||0.295||
 cpc_codes_mem|stem|5fcef0a36b7afca8|ed8acfe423c0a853|485f8713c163dbcd|-1726884630||t19968.7761941308s|8c66a877d5351e9e|881|rds|local|vector|||0.001||
 cpc_raw|stem|a1cfc330de9c55ac|84edb6fd9a718ad7|5927f53d1a43fd9b|1538963313||t19968.6457792864s|d684a3f50d50d6f2|532625286|rds|local|vector|||10.251||
 dl_me_raw_stuff|function|bbeea75760693d15|||||||||||||||
@@ -16,7 +16,7 @@ get_me_patent_assignee_loc|function|9f9a56083cdd46ff|||||||||||||||
 get_me_us_counties|function|ea118b9fb2196670|||||||||||||||
 get_me_us_counties2010|function|7f525f126121ff29|||||||||||||||
 get_me_us_location|function|db531d5101ab8c23|||||||||||||||
-get_patent_counts_wide|function|0f59de5c81d6af22|||||||||||||||
+get_patent_counts_wide|function|734e51ad0930f668|||||||||||||||
 get_rel_table_co_year|function|d287b52346ee8b81|||||||||||||||
 inventor_file|stem|26a7ddc8c95863d0|d9e718d29ceaae27|2c530c1562a7fbd1|1314243806|data/data_raw/unzipped/g_inventor_disambiguated.tsv|t19941.7788223155s|e56307200b8a22cc|2132722674|file|local|vector|||0.243||
 inventor_raw|stem|0e2ae26d32a9cc15|fba29481ab30dd98|93fc7708ae1da104|-1009114140||t19965.6097131911s|18d8ecd1d3c21eca|606647424|rds|local|vector|||26.313||
@@ -25,10 +25,12 @@ location_raw|stem|802d6555cd4e88c5|047650cd5c6838a9|b09db675b4eb6e7f|1531131639|
 location_us|stem|dffc8b36abb2b591|ed35c5e31ba17e65|22c44b4525bc1ae8|-997291703||t19965.6105658674s|bbd488e0b2866652|245658|rds|local|vector|||0.011||
 my_unique|function|1f5a0b56fd2d4ce2|||||||||||||||
 patent_assignee_location|stem|3770aabe50951e6e|eaa8c84419f6f270|60d26c67510f2fb5|367768944||t19965.6124869727s|8855194edf927d50|721339552|rds|local|vector|||56.99||
-patent_counts_wide|stem|3a0ed46c13e8057b|417f1ae42f034904|ecd657167b003330|1702443325||t19968.8406675585s|1964ef0873285cf5|22939243|rds|local|vector|||39.341||
+patent_counts_wide|stem|f20cc0f70ae74d99|417f1ae42f034904|a9ed78f9bb1957e5|1702443325||t19969.828204981s|0330683901cbd0f2|7543694|rds|local|vector|||40.57||
+patent_db|stem|055c5f99272a555e|e63e11c71db485ef|db1c640ab14bc3aa|-242810428||t19969.8360041808s|ba41eabafff2540f|48|rds|local|vector|||8.287||
 patent_file|stem|6382176349c156c2|784985d25b1ff5b8|2c530c1562a7fbd1|361076240|data/data_raw/unzipped/g_patent.tsv|t19941.7791886955s|fe68d5009b09dd52|1055731960|file|local|vector|||0||
 patent_raw|stem|59f0629f2402835a|6a600ad9163aa470|81c1fde921865d1e|-730994296||t19965.6101543309s|bd06401e3a9409f8|202280349|rds|local|vector|||13.25||
 read_sample|function|50fa3240879e5a1a|||||||||||||||
 rel_geoid_year|stem|8b27de405741d886|9be31d3b0119a218|60401893c8bf23c7|1133841084||t19968.58271356s|131af36b92dd4e15|477311|rds|local|vector|||0.638||
 slim_cpc|function|24c66946651495b3|||||||||||||||
 table_with_options|function|8aaa3df01d5209e3|||||||||||||||
+write_to_proj_erc|function|bfd8d7837a48bec9|||||||||||||||