如何在R语言中基于另一类别对连续日期分组（附示例数据）

Got it, let's solve this problem where you need to group consecutive date sequences within each member category in your trainall data frame. Here are two straightforward approaches—one using the tidyverse (dplyr) and another with base R, so you can pick what works best for you.

Step 1: Prepare the Data (Critical First Step!) #

First, we need to convert your character-based time column into a proper Date format. This is essential for calculating date differences accurately:

# Convert time to Date type (matches dd/mm/yyyy format of your data)
trainall$time <- as.Date(trainall$time, format = "%d/%m/%Y")

Approach 1: Using dplyr (Tidyverse) #

This method is intuitive and readable, great for those who prefer the tidyverse workflow:

1. First, load the dplyr package (install it with install.packages("dplyr") if you haven't already):

library(dplyr)

2. Sort the data, group by member, and calculate consecutive date groups:

trainall_grouped <- trainall %>%
  # Sort data by member and date to ensure sequence order
  arrange(member, time) %>%
  # Group by each member to analyze their dates separately
  group_by(member) %>%
  # Calculate the difference between current date and the previous date
  mutate(date_gap = time - lag(time, default = first(time))) %>%
  # Create a group ID: increment whenever the date gap is more than 1 day
  mutate(group_id = cumsum(date_gap > 1)) %>%
  # Optional: remove grouping if you don't need it anymore
  ungroup()

What this does: #

• arrange(member, time) makes sure dates are in chronological order for each member.
• date_gap measures how many days separate each date from the prior one. The first date uses itself as the "previous" date, so its gap is 0.
• cumsum(date_gap > 1) adds 1 to the group ID every time we hit a gap larger than 1 day, creating a unique ID for each consecutive date block.

Approach 2: Using Base R #

If you prefer not to load extra packages, this base R method works just as well:

# First, sort the data by member and date
trainall <- trainall[order(trainall$member, trainall$time), ]

# Create group IDs using ave() to calculate per member
trainall$group_id <- ave(
  x = as.numeric(trainall$time),  # Convert dates to numeric for difference calculation
  by = trainall$member,
  FUN = function(x) {
    # Calculate gaps between consecutive dates, mark gaps >1 day
    gaps <- c(1, diff(x) > 1)
    # Cumulative sum to assign group IDs
    cumsum(gaps)
  }
)

How this works: #

• as.numeric(trainall$time) converts Date objects to their underlying numeric values (days since 1970-01-01), making it easy to calculate differences.
• diff(x) > 1 identifies where the date gap is more than one day. We prepend 1 to handle the first date (which starts a new group).
• cumsum(gaps) builds the group ID by counting each time we hit a new non-consecutive date.

Example Output #

After running either method, your modified trainall (or trainall_grouped) data frame will have a group_id column that groups consecutive dates per member. For example:

time member group_id
1 1951-01-01      1        0
2 1951-01-02      1        0
3 1951-01-03      1        0
4 1951-01-04      1        0
5 1953-03-03      1        1
6 1953-03-04      1        1
...

Here, member 1 has two distinct date groups: the January 1951 sequence (group 0) and the March 1953 sequence (group 1).

内容的提问来源于stack exchange，提问作者Judith