如何在R中依据条件数据框修改目标数据列的值？

Your loop-based solution works, but a more idiomatic and efficient way to handle this in the tidyverse is using a non-equi join to match each timestamp in data_df to the corresponding range in criteria_df, then updating the values accordingly. This avoids explicit loops and leverages vectorized operations, which are faster for larger datasets.

Here's how to implement it:

library(tidyverse)

# Original data
set.seed(123)
data_df <- tibble(t = 1:15, value = sample(letters, 15))

# Criteria data
criteria_df <- tibble(start = c(1, 3, 7), end = c(2, 5, 10), value = c('a', 'b', 'c'))

# Update values using non-equi join
updated_data <- data_df %>%
  # Join data_df with criteria_df where t falls between start and end
  left_join(
    criteria_df,
    join_by(between(t, start, end)),
    suffix = c("_original", "_new")
  ) %>%
  # Use the new value if available, else keep original
  mutate(value = coalesce(value_new, value_original)) %>%
  # Keep only the necessary columns
  select(t, value)

# View the result
updated_data

How This Works: #

1. Non-Equi Join: The join_by(between(t, start, end)) condition matches each t in data_df to the row(s) in criteria_df where t lies within the start-end range. For timestamps not in any range, the value_new column will be NA.
2. Coalesce: coalesce(value_new, value_original) picks the first non-NA value, which means we use the criteria-specified value when available, otherwise retain the original value.
3. Cleanup: We select only the t and updated value columns to get back the original dataframe structure.

Handling Overlapping Ranges (If Applicable): #

If your criteria_df has overlapping ranges, the join will create multiple rows for timestamps that fall into multiple ranges. In that case, you can add a step to prioritize which value to use (e.g., the latest range, or the first match). For example, if you want to keep the last matching criteria:

updated_data <- data_df %>%
  left_join(criteria_df, join_by(between(t, start, end)), suffix = c("_original", "_new")) %>%
  group_by(t) %>%
  slice_last() %>% # Keep the last matching criteria row
  ungroup() %>%
  mutate(value = coalesce(value_new, value_original)) %>%
  select(t, value)

Alternative: Functional Approach with purrr #

If you prefer a functional style over joins, you can use purrr::reduce to iteratively apply each criteria range to the dataframe. This is similar to your loop but more concise:

library(purrr)

updated_data <- criteria_df %>%
  pmap(function(start, end, value) {
    data_df %>% mutate(value = if_else(between(t, start, end), value, value))
  }) %>%
  reduce(full_join, by = c("t", "value")) %>%
  distinct(t, .keep_all = TRUE)

However, this approach is less efficient than the non-equi join for large datasets, as it involves multiple dataframe operations.

Content of the question originates from Stack Exchange, asked by mauna