priceR (version 0.1.0)

extract_salary: Extract numeric salary from text data

Description

Extract numeric salary from text data. `extract_salary` automatically converts weekly and hourly rates to amounts per annum.

Usage

extract_salary(salary_text, exclude_below, exclude_above, salary_range_handling,
include_periodicity, hours_per_workday, days_per_workweek, working_weeks_per_year)

Arguments

salary_text

A character string, or vector of character strings.

exclude_below

A lower bound. Anything lower than this number will be replaced with NA.

exclude_above

An upper bound. Anything above this number will be replaced with NA.

salary_range_handling

A method of handling salary ranges. Defaults to returning an average of the range; can also be set to "max" or "min".

include_periodicity

Set to TRUE to return an additional column stating the detected peridicity in the character string.

hours_per_workday

Set assumed number of hours in the workday. Only affects annualisation of rates indentified as Daily. Default is 8 hours.

days_per_workweek

Set assumed number of days per workweek. Only affects annualisation of rates indentified as Daily. Default is 5 days.

working_weeks_per_year

Set assumed number of working weeks in the year. Only affects annualisation of rates indentified as Daily or Weekly. Default is 48 weeks.

Examples

# NOT RUN {
# Provide a salary string and 'extract_salary' and will extract the salary and return it
extract_salary("$160,000 per annum")
# 160000


# If a range is present, the average will be taken by default
extract_salary("$160,000 - $180000.00 per annum")
# 170000


# Take the 'min' or 'max' of a salary range by setting salary_range_handling parameter accordingly
extract_salary("$160,000 - $180000.00 per annum", salary_range_handling = "min")
# 160000


# Extract salaries from character string(s)
annual_salaries <- c("$160,000 - $180000.00 per annum",
                     "$160000.00 - $180000.00 per annum",
                     "$145000 - $155000.00 per annum",
                     "$70000.00 - $90000 per annum",
                     "$70000.00 - $90000.00 per annum plus 15.4% super",
                     "$80000.00 per annum plus 15.4% super",
                     "60,000 - 80,000",
                     "$78,686 to $89,463 pa, plus 15.4% superannuation",
                     "80k - 100k")

extract_salary(annual_salaries)
# 170000 170000 150000  80000  53338  40008  70000  56055  90000


# Automatically detect, extract, and annualise daily rates
daily_rates <- c("$200 daily", "$400 - $600 per day", "Day rate negotiable dependent on experience")
extract_salary(daily_rates)
# 48000 120000     NA


# Automatically detect, extract, and annualise hourly rates
hourly_rates <- c("$80 - $100+ per hour", "APS6/EL1 hourly rate contract")
extract_salary(hourly_rates)
# 172800   6720
# Note 6720 is undesirable. Setting the lower and upper bounds sensibly avoids this


salaries <- c(annual_salaries, daily_rates, hourly_rates)


# Setting lower and upper bounds provides a catch-all to remove unrealistic results
# Out of bounds values will be converted to NA
extract_salary(salaries, exclude_below = 20000, exclude_above = 600000)
# 170000 170000 150000  80000  53338  40008  70000  56055  90000  48000 120000     NA 172800     NA


# extract_salary automatically annualises hourly and daily rates
# It does so by making assumptions about the number of working weeks in a year,
# days per workweek, and hours per workday
# And the assumed number of hours per workday can be changed from the default (8)
# The assumed number of workdays per workweek can be changed from the default (5)
# The assumed number of working weeks in year can be changed from the default (48)
# E.g.
extract_salary(salaries, hours_per_workday = 7, days_per_workweek = 4, working_weeks_per_year = 46)
# 170000 170000 150000  80000  53338  40008  70000  56055  90000  36800  92000     NA 115920   4508


# To see which salaries were detected as hourly or weekly, set include_periodicity to TRUE
extract_salary(salaries, include_periodicity = TRUE)

# salary periodicity
# 1  170000      Annual
# 2  170000      Annual
# 3  150000      Annual
# 4   80000      Annual
# 5   53338      Annual
# 6   40008      Annual
# 7   70000      Annual
# 8   56055      Annual
# 9   90000      Annual
# 10  48000       Daily
# 11 120000       Daily
# 12     NA       Daily
# 13 172800      Hourly
# 14   6720      Hourly


# }