使用dplyr计算非统一采样频率的月度叶片生长率问题

Great question! Mixed sampling frequencies (like weekly vs. bi-monthly) can throw a wrench into straightforward lag-based calculations, but your intuition to standardize the data first is spot-on. Here’s a step-by-step solution using dplyr:

1. First: Standardize to Monthly Frequency #

The core issue here is that your raw data has uneven row counts per month. To calculate a meaningful month-over-month growth rate, you first need to collapse each month’s data into a single representative value for each plant. You have two common options for this representative value (pick based on your research goals):

• Monthly average: Smooths out sampling noise if your measurements have variability
• Final monthly measurement: Better if you want to capture the end-of-month growth state

Code to Standardize the Data #

library(dplyr)

# Start with your mixed-frequency dataset
toy_growthrate_with_twist <- data.frame(
  Plant_ID = c("365","365","365","365","365","365","365","365","365","365","365","365"),
  Leaf_length = c(1,2,4, 10, 15, 17, 20, 25, 30, 50, 45, 47),
  Month = c(5,5,5,5,6,6,7,7,8,8,9,9),
  Period = c("T1","T2","T3","T4","T1","T2","T1","T2","T1","T2","T1","T2")
)

# Group by plant and month, then calculate representative values
monthly_standardized <- toy_growthrate_with_twist %>%
  group_by(Plant_ID, Month) %>%
  summarise(
    avg_leaf_length = mean(Leaf_length),  # Monthly average
    final_leaf_length = last(Leaf_length) # Final measurement of the month
  ) %>%
  ungroup()

2. Calculate Monthly Growth Rates #

Now that you have one row per month per plant, you can use lag() safely to compare each month to the previous one:

# Arrange data by plant and month, then compute growth rates
monthly_growth <- monthly_standardized %>%
  arrange(Plant_ID, Month) %>%
  mutate(
    # Growth rate using monthly average
    growth_pct_avg = ((avg_leaf_length - lag(avg_leaf_length)) / lag(avg_leaf_length)) * 100,
    # Growth rate using final monthly measurement
    growth_pct_final = ((final_leaf_length - lag(final_leaf_length)) / lag(final_leaf_length)) * 100
  )

# View the result
monthly_growth

Why This Works #

• By aggregating to a monthly frequency first, you eliminate the mismatch in row counts between months (4 rows for May vs. 2 for June, etc.). This ensures lag() always pulls the entire previous month’s representative value, not a random earlier sampling point.
• Choosing between average or final measurement depends on your study’s context: if leaf growth is steady, final measurement is more reflective of monthly progress; if measurements have day-to-day variation, average will smooth out noise.

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