match_2_bank: The match_2_bank() function matches elements to a bank of...

In mikejacktzen/datzen:

Description

Elements in df_raw[,col_name] will be fuzzyjoined to elements in bank_entity

Usage

match_2_bank(bank_entity, col_name, df_raw, method = "jw", max_dist = 0.2)

Arguments

bank_entity	a single column data.frame that contains the unique population of entities inside bank_entity$bank_entity
col_name	a single string for the name of the column in 'df_raw' to be processed and matched to bank_entity
df_raw	a data.frame that contains the column specified by the 'col_name' argument
method	the method argument passed to stringdist_left_join
max_dist	the max_dist argument passed to stringdist_left_join

Value

A 2 column data.frame with elements in df_raw[,col_name] fuzzyjoined to 'bank_entity' via stringdist_left_join

Examples

#######################################
# Create name bank of unique names
# all population of names
# need to disambiguate, get rid of redundant
# most likely manually
#######################################
library(dplyr)


df_raw = data.frame(name_a=c('joseph','joe','jo ',' joesephine','joseph','jo'),
                    name_b=c('joseph','jo3','jo ',' joesephine','joseph','jo'),
                    name_c=c('joseph','j0 '))

names_col_to_pool = c('name_a','name_b','name_c')
name_all_unique = as.character(unlist(df_raw[,names_col_to_pool])) %>% unique %>% sort

# creating name bank: drop mispelled, appeal to appx string match latter
bank_entity = name_all_unique %>% clean_name() %>% unique %>% sort %>%
  gsub(.,pattern='^(Jo3)$|(J0)|(j0)|(jo3)',replace='') %>% # Joseph
  # spotcheck additional manual gsubs()
  unique %>% sort

bank_entity = data.frame(bank_entity=bank_entity)
str(bank_entity)


library(stringdist)

# run one example to figure out distance boundary
stringdist::stringdist("Jo","Joseph",method='jw')
# c("osa", "lv","dl", "hamming", "lcs", "qgram", "cosine", "jaccard", "jw", "soundex")

match_2_bank(bank_entity=bank_entity,col_name="name_a",df_raw=df_raw)

match_2_bank(bank_entity=bank_entity,
             col_name="name_b",df_raw=df_raw,
             max_dist=0.6,method='jw')


# make sure output of match_2_bank() == nrow(df_raw)
nrow(df_raw)

##############################
# Replace
##############################
# head(df_raw)
#
# dat_edit = df_raw
# dat_edit[,'name_a'] = match_2_bank(bank_entity=bank_entity,col_name="name_a",df_raw=df_raw,max_dist=0.2,method='jw')[,2]
# dat_edit[,'name_b'] = match_2_bank(bank_entity=bank_entity,col_name="name_b",df_raw=df_raw,max_dist=0.2,method='jw')[,2]
# dat_edit[,'name_c'] = match_2_bank(bank_entity=bank_entity,col_name="name_c",df_raw=df_raw,max_dist=0.2,method='jw')[,2]
#
# head(dat_edit)
#
#' # dat_edit[,c('name_a','name_b','name_c')]
# df_raw[,c('name_a','name_b','name_c')]

mikejacktzen/datzen documentation built on June 14, 2019, 5:23 p.m.