如何在R语言数据框中按特定模式合并语音时长数据

Got it, let's work through this problem step by step. You need to merge durations when "n't" follows vowel-ending words like "do" or "ca", while leaving consonant-ending pairs (like "did n't") untouched. Here's a straightforward approach using R:

Step 1: Define a Row Processing Function #

First, we'll create a function to handle each row individually. This lets us check for "n't" occurrences, validate the preceding word, and merge durations/words as needed:

process_row <- function(row) {
  # Extract duration and word columns from the row
  dur_cols <- paste0("d", 1:10)
  word_cols <- paste0("w", 1:10)
  durations <- row[dur_cols]
  words <- row[word_cols]
  
  # Find all positions where the word is "n't"
  nt_positions <- which(words == "n't")
  
  # Iterate backwards to avoid position shifts messing up later checks
  for (pos in rev(nt_positions)) {
    if (pos > 1) {
      prev_word <- words[pos - 1]
      # Check if preceding word is "do" or "ca" (vowel-ending targets)
      if (prev_word %in% c("do", "ca")) {
        # Merge the two words into a single contraction
        words[pos - 1] <- paste0(prev_word, "n't")
        # Sum their durations
        durations[pos - 1] <- durations[pos - 1] + durations[pos]
        # Remove the standalone "n't" column, shift remaining columns left
        words <- words[-pos]
        durations <- durations[-pos]
        # Fill empty spots with NA to keep 10 columns total
        if (length(words) < 10) {
          words <- c(words, rep(NA, 10 - length(words)))
        }
        if (length(durations) < 10) {
          durations <- c(durations, rep(NA, 10 - length(durations)))
        }
      }
    }
  }
  
  # Reconstruct the processed row with original column names
  new_row <- c(durations, words)
  names(new_row) <- c(dur_cols, word_cols)
  return(new_row)
}

Step 2: Apply the Function to Your DataFrame #

Now we'll use apply() to run this function on every row of test1, then convert the result back to a dataframe:

# Process all rows
test_processed <- as.data.frame(t(apply(test1, 1, process_row)))

# Restore original row names
rownames(test_processed) <- rownames(test1)

# Ensure duration columns are numeric (they get coerced to character during transpose)
test_processed[, paste0("d", 1:10)] <- lapply(test_processed[, paste0("d", 1:10)], as.numeric)

Step 3: Check the Results #

If you print test_processed, you'll see exactly what you're looking for:

• Row 10: "do" and "n't" become "don't" with a merged duration of 0.064 + 0.075 = 0.139
• Row 132: "ca" and "n't" become "can't" with a merged duration of 0.087 + 0.088 = 0.175
• Row 784: "did n't" pairs stay separate since "did" ends with a consonant

Key Notes #

• We iterate backwards through "n't" positions so that shifting columns left doesn't break the position indices for remaining checks.
• The function keeps the dataframe structure (10 duration + 10 word columns) by filling empty spots with NA after merging.
• This approach handles multiple "n't" occurrences in a single row automatically.

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