数据插补后如何将因子列q1取整至最近的因子水平

Hey there! Let's tackle your problem of rounding the q1 values to the nearest specified factor levels (0, 20, 40, 60, 80, 100) in your dataframe. You mentioned needing to handle values like 49 and 91 specifically, so here are a couple of straightforward R solutions that'll get this done:

Method 1: Using findInterval for Fast Interval Mapping #

This method uses base R's findInterval to quickly assign each value to the closest target level. We'll add a small offset to ensure midpoints (if any) are handled correctly, but you can adjust this based on your rounding preference:

# Define your target factor levels
target_levels <- c(0, 20, 40, 60, 80, 100)

# Round q1 to the nearest target level
df$q1_rounded <- target_levels[findInterval(df$q1 + 10, target_levels)]
# The +10 ensures values exactly halfway between two levels round up; remove or adjust if you want different behavior

# Convert the rounded column to factor type (matching your target levels)
df$q1_rounded <- factor(df$q1_rounded, levels = target_levels)

What this does: #

• 49 will round to 40 (since it's 9 units away from 40, vs. 11 units from 60)
• 91 will round to 100 (9 units away from 100, vs. 11 units from 80)

Method 2: Using purrr for Explicit Distance Calculation #

If you prefer a more intuitive approach where we calculate the distance from each q1 value to all target levels, the purrr package makes this easy:

library(purrr)

target_levels <- c(0, 20, 40, 60, 80, 100)

# For each value in q1, find the target level with the smallest distance
df$q1_rounded <- map_dbl(df$q1, function(x) {
  distance_to_levels <- abs(x - target_levels)
  target_levels[which.min(distance_to_levels)]
})

# Convert to factor
df$q1_rounded <- factor(df$q1_rounded, levels = target_levels)

This method is more readable if you want to explicitly see how the closest level is determined, and it gives the same result as Method 1 for your specific values.

Final Result #

After running either method, your dataframe will have a new q1_rounded column (or you can overwrite the original q1 column if needed) with the rounded, factor-type values. Here's what the output looks like:

id visit trt q1 q1_rounded
1  300    1   0  0          0
2  300    2   A 100        100
3  300    3   0  0          0
4  300    4   B 49         40
5  300    5   0  0          0
6  300    6   C 60         60
7  450    1   0  0          0
8  450    2   A 91        100
9  450    3   0  0          0
10 450    4   B 40         40
11 450    5   0  0          0
12 450    6   C 20         20

If you want to replace the original q1 column directly, just replace q1_rounded with q1 in the code.

内容的提问来源于stack exchange，提问作者user11916948