您可以使用cumsum(c(1, diff(day) != 1))生成连续日期组。

library(dplyr)

df %>%
  group_by(year) %>%
  mutate(num.hw = cumsum(c(1, diff(day) != 1))) %>%
  group_by(year, num.hw) %>%
  summarise(length.hw = n(), avg.temp = mean(temp), .groups = 'drop')

# # A tibble: 4 × 4
#    year num.hw length.hw avg.temp
#   <int>  <dbl>     <int>    <dbl>
# 1  2011      1         3     96.3
# 2  2011      2         4     98.5
# 3  2012      1         4     96.2
# 4  2012      2         3     99.3

数据

df <- read.table(text = "
  year  day  temp
  2011  150  96 
  2011  151  96
  2011  152  97
  2011  166  98
  2011  167  99
  2011  168  98
  2011  169  99
  2012  177  96
  2012  178  96
  2012  179  95
  2012  180  98
  2012  195  100
  2012  196  99
  2012  197  99", header = TRUE)

数据表解决方案

与Darren Tsai的方法完全相同，他提出了一个很好的累积和方法。

library(data.table)

setDT(df)

df[, num.hw := cumsum(c(1, diff(day) != 1)), year]
df[, .(length.hw = .N, avg.temp = mean(temp)), by = .(year, num.hw)]

# year num.hw length.hw avg.temp
# 1: 2011      1         3   96.333
# 2: 2011      2         4   98.500
# 3: 2012      1         4   96.250
# 4: 2012      2         3   99.333