コミット: 569fd3539a771af93df5e82d49c77b29f19bf175 - myprojects-hg-reborn (hg)

コミットメタ情報

• リビジョン: 569fd3539a771af93df5e82d49c77b29f19bf175 (tree)
• 日時: 2022-09-29 23:19:28
• 作者: Lorenzo Isella <lorenzo.isella@gmai...>
• コミッター: Lorenzo Isella

ログメッセージ

I added a script to look (in parallel) for multiple keywords and showing the progress of the search.

変更サマリ

• add: R-codes/search_tam_scoreboard.R

差分

diff -r 27289eb79302 -r 569fd3539a77 R-codes/search_tam_scoreboard.R
--- /dev/null	Thu Jan 01 00:00:00 1970 +0000
+++ b/R-codes/search_tam_scoreboard.R	Thu Sep 29 16:19:28 2022 +0200
@@ -0,0 +1,80 @@
+rm(list=ls())
+
+library(tidyverse)
+library(openxlsx)
+library(janitor)
+library(furrr)
+
+source("/home/lorenzo/myprojects-hg/R-codes/stat_lib.R")
+
+
+find_text_multiple_keywords_exact_matches_future <- function( df, keywords){
+
+    res <- future_map_dfr(keywords,
+                         function(x) find_text_filter_exact_matches(df, x) ,
+                         .progress =T)
+
+    return(res)
+    
+}
+
+
+
+#################################################################à
+n_cores <- 2
+
+plan(multicore(workers=return_cores(n_cores)))
+
+
+
+df_sc_ini <- readRDS("scoreboard.RDS")
+
+df_sc <- df_sc_ini |>
+    all_to_lower()
+
+rm(df_sc_ini)
+
+df_tam_ini <- readRDS("TAM_cleaned_for_shiny.RDS")
+
+df_tam <- df_tam_ini |>
+    all_to_lower()
+
+rm(df_tam_ini)
+
+keywords <- read_delim("words.txt", delim="\n", col_names=F) |>
+    clean_data() |>
+    all_to_lower() |>
+    ## mutate(x1=paste("(?i)", x1, sep="")) |> 
+    pull(x1)
+
+print("Done reading the data")
+
+tam_words <- find_text_multiple_keywords_exact_matches_future(df_tam, keywords)
+
+
+save_excel(tam_words, "tam_results.xlsx")
+
+
+tam_words_unique <- tam_words |>
+    distinct(case_reference, .keep_all=T)
+
+
+save_excel(tam_words_unique, "tam_results_short.xlsx")
+
+
+print("Done reading the TAM")
+
+
+scoreboard_words <- find_text_multiple_keywords_exact_matches_future(df_sc, keywords)
+
+
+save_excel(scoreboard_words, "scoreboard_results.xlsx")
+
+
+scoreboard_words_unique <- scoreboard_words |>
+    distinct(case_number, .keep_all=T)
+
+save_excel(scoreboard_words_unique, "scoreboard_results_short.xlsx")
+
+
+print("So far so good")