Using for loops, while, tidyverse, or packages to create a dataset with matching characteristics from a previous one (sampling)
I'm working with panel data. We assessed children in 2019 and 2020. Therefore, I have two datasets (2019 and 2020) and I want to create a third dataset matching the data from the second dataset (2020) that matches the characteristics of the first dataset (2019).This third dataset will have fewer participants, but they'll share the same characteristics of their "peers" from 2019. So, the proportion of boys and girls will be about the same of 2019, the mother's age will be about the same, etc.
Examaple:
Code:
df_2019 = structure(list(asqse_quest = c(24, 24, 24, 24, 24, 24, 24, 24,
24, 24, 24, 24, 24, 24, 24, 24, 24, 24, 24, 24, 24, 24, 24, 24,
24, 24, 24, 24, 24, 24, 24, 24, 24, 24, 24, 24, 24, 24, 24, 24,
24, 24, 24, 24, 24, 24, 24, 24, 24, 24), year_completed_cat = structure(c(2L,
2L, 2L, 2L, 2L, 2L, 2L, 2L, 2L, 2L, 2L, 2L, 2L, 2L, 2L, 2L, 2L,
2L, 2L, 2L, 2L, 2L, 2L, 2L, 2L, 2L, 2L, 2L, 2L, 2L, 2L, 2L, 2L,
2L, 2L, 2L, 2L, 2L, 2L, 2L, 2L, 2L, 2L, 2L, 2L, 2L, 2L, 2L, 2L,
2L), levels = c("18", "19", "20", "21", "22", "23", "24"), class = "factor"),
sex_male = c(1, 0, 0, 1, 1, 1, 1, 0, 1, 0, 1, 1, 1, 0, 0,
1, 0, 0, 1, 0, 1, 1, 1, 0, 1, 1, 1, 1, 0, 1, 1, 0, 1, 1,
0, 1, 1, 1, 0, 1, 1, 0, 1, 0, 1, 1, 1, 0, 1, 0), momage = c(36,
39, 22, 20, 29, 40, 31, 37, 29, 38, 24, 35, 32, 30, 32, 31,
29, 21, 28, 29, 40, 21, 38, 29, 28, 33, 25, 25, 30, 29, 25,
27, 28, 31, 24, 28, 35, 29, 17, 35, 32, 29, 27, 24, 29, 25,
28, 24, 21, 26), momed = c(4, 4, 2, 2, 4, 3, 2, 3, 2, 4,
3, 4, 4, 4, 4, 4, 3, 4, 3, 4, 4, 2, 2, 4, 4, 4, 4, 4, 4,
4, 2, 4, 3, 3, 3, 3, 4, 4, 2, 4, 4, 3, 2, 2, 3, 4, 4, 3,
2, 4), income = c(4, 4, 2, 3, 4, 1, 2, 5, 4, 4, 5, 4, 4,
4, 4, 4, 4, 2, 3, 3, 4, 2, 3, 4, 4, 4, 5, 4, 3, 3, 4, 4,
3, 4, 1, 4, 2, 4, 3, 4, 4, 3, 4, 3, 4, 4, 4, 3, 4, 4)), class = "data.frame", row.names = c(NA,
-50L))
df_2020 = structure(list(asqse_quest = c(24, 24, 24, 24, 24, 24, 24, 24,
24, 24, 24, 24, 24, 24, 24, 24, 24, 24, 24, 24, 24, 24, 24, 24,
24, 24, 24, 24, 24, 24, 24, 24, 24, 24, 24, 24, 24, 24, 24, 24,
24, 24, 24, 24, 24, 24, 24, 24, 24, 24), year_completed_cat = structure(c(3L,
3L, 3L, 3L, 3L, 3L, 3L, 3L, 3L, 3L, 3L, 3L, 3L, 3L, 3L, 3L, 3L,
3L, 3L, 3L, 3L, 3L, 3L, 3L, 3L, 3L, 3L, 3L, 3L, 3L, 3L, 3L, 3L,
3L, 3L, 3L, 3L, 3L, 3L, 3L, 3L, 3L, 3L, 3L, 3L, 3L, 3L, 3L, 3L,
3L), levels = c("18", "19", "20", "21", "22", "23", "24"), class = "factor"),
sex_male = c(1, 1, 1, 1, 0, 1, 0, 1, 0, 1, 0, 1, 1, 0, 1,
0, 0, 1, 1, 0, 1, 0, 1, 0, 0, 1, 1, 0, 1, 1, 1, 0, 1, 0,
1, 1, 1, 1, 1, 1, 1, 1, 1, 0, 1, 1, 0, 1, 1, 1), momage = c(23,
26, 33, 34, 29, 26, 23, 29, 40, 36, 33, 18, 31, 31, 31, 32,
34, 35, 29, 37, 19, 30, 33, 25, 32, 35, 37, 27, 23, 29, 28,
26, 30, 27, 38, 28, 29, 39, 26, 25, 29, 39, 35, 32, 20, 38,
31, 27, 28, 23), momed = c(2, 4, 4, 3, 4, 3, 2, 2, 3, 4,
1, 2, 2, 4, 4, 4, 4, 2, 4, 4, 2, 4, 4, 4, 2, 4, 4, 2, 4,
2, 1, 4, 3, 2, 4, 4, 4, 2, 4, 2, 4, 4, 4, 4, 2, 4, 4, 4,
4, 1), income = c(2, 4, 4, 4, 4, 5, 3, 2, 2, 4, 1, 3, 4,
5, 1, 4, 3, 1, 4, 5, 5, 4, 4, 4, 3, 4, 4, 2, 4, 5, 1, 4,
4, 1, 4, 4, 4, 4, 3, 4, 4, 4, 5, 4, 2, 4, 4, 4, 4, 4)), class = "data.frame", row.names = c(NA,
-50L))
Created on 2024-07-12 with reprex v2.1.0
You can try the MatchIt package, which has a function that performs propensity score matching.
We first combine the two datasets with bind_rows, assigning an id to distinguish the two datasets:
library(dplyr)
data <- bind_rows(df_2019, df_2020, .id="year") |>
mutate(year=+(year==1)) # 1=2019 (treated), 0=2020 (controls)
Rows corresponding to year==1 are your treated (from 2019 data) and year==0 corresponds to your controls (from 2020 data).
To find controls that match as closely to the treated as possible, we can use the matchit function. There are a number of arguments, and for brevity, we'll just use the defaults.
library(MatchIt)
We'll first try to match exactly on years completed, sex, and mother's age and see if we have any luck.
match_obj <- matchit(year ~ asqse_quest+year_completed_cat+sex_male+momage+momed+income,
data = data,
exact= ~ year_completed_cat+sex_male+momage,
replace = FALSE)
#Error in `matchit()`:
#! No matches were found.
No surprise because the two datasets don't match at all on years completed. Let's make our matching condition less strict
match_obj <- matchit(year ~ asqse_quest+year_completed_cat+sex_male+momage+momed+income,
data = data,
exact= ~ sex_male+momage,
replace = FALSE)
There's no error this time, but we get a warning
#Warning message:
#Fewer control units than treated units in some `exact` strata; not all treated units will get a match.
That's ok. Now summarise the results.
summary(match_obj)
...
Sample Sizes:
Control Treated
All 50 50
Matched 25 25
Unmatched 25 25
Discarded 0 0
The output says that we found 25 controls from the original 50. Other useful information is given but I've omitted here for brevity. Now use match.data to get the matches, together with the original treated.
matched_data <- match.data(match_obj)
Now we simply filter out the treated and we are left with the matching controls:
df_2020_new <- filter(matched_data, year==0)
head(df_2020_new)
asqse_quest year_completed_cat sex_male momage momed income
1 24 20 1 23 2 2
2 24 20 1 26 4 4
3 24 20 1 33 4 4
4 24 20 1 34 3 4
5 24 20 0 29 4 4
6 24 20 1 26 3 5
7 24 20 0 23 2 3
8 24 20 1 29 2 2
9 24 20 0 40 3 2
10 24 20 1 36 4 4
Check the help page for matchit to see how you can modify the methods for matching. There's too much detail to give here, but this is the basic idea.
- cbind multiple data.frames in a list with unequal lengths
- How to make looping to be the fastest in R?
- Not able to to convert R data frame to Spark DataFrame
- Scraping fund meta data in Yahoo Finance (not prices)
- How do I pivot_wider() so that duplicates are maintained in their own columns?
- geom_tile unable to sort - diagonal
- Using timeDiff() to work out the difference between times and selecting ones that are over an hour out
- Add vertical line in a DT datatable in a shiny app with custom container
- "cannot coerce type 'closure' to vector of type 'character'" error when trying to convert date column
- R and writing to JSON
- How can I display the group mean value and total mean value together in a bar chart using ggplot?
- How to customize nticks in persp R
- Combine likert chart and bar plot ordered by the same order in R using ggplot2
- Format date as Year/Quarter
- How to create a swimmer plot with two independent axises which the primary one representing follow-up length, while the secondary being calendar time
- How to indent a heading under a numbered list in Rmarkdown knitting to HTML?
- Widen/narrow line sections of plot so that x-axis breaks at peaks are evenly spaced
- Mutate across multiple columns using dplyr
- Get dataframe from function into workspace
- Check whether values in one data frame column exist in a second data frame
- Nested while loop behavior in R
- Recursively Converting a Data Frame to a Nested List Where the Level of Nestedness of the List Equals the Number of Columns in the Data Frame
- Split data.frame based on levels of a factor into new data.frames
- How to create a new column using complex fomulas and variable outputs (i.e., recoding variables) in a Pandas DataFrame similarly to R's `case_when`?
- How to increase the font size of ggtitle in ggplot2
- AIC or BIC calculation in R?
- Dynamic interaction between mutate and ud.convert leading to miscalculation
- Calculate the mean for date and hour from a multi year dataframe in r
- Sorting output from variable importance (caret package)
- R: Manually change the color of stat_pvalue_manual?