基于参考值从下往上计算DataFrame新列的实现方法

我们可以通过两种常用方法实现反向计算生成relToDip列：

方法一：Base R 循环实现 #

逻辑直观，直接按照从下往上的顺序逐行计算：

# 加载数据
df <- structure(list(Rec = 1:10, DateTime = structure(c(1585468800, 
1585472400, 1585476000, 1585479600, 1585483200, 1585486800, 1585490400, 
1585494000, 1585497600, 1585501200), class = c("POSIXct", "POSIXt"
), tzone = "GMT"), TempHMB5 = c(6.78, 6.78, 6.78, 6.78, 6.78, 
6.77, 6.77, 6.77, 6.77, 6.76), PressHMB5 = c(1074.09, 1074.87, 
1074.8, 1075.02, 1074.95, 1074.95, 1074.59, 1074.45, 1074.45, 
1074.52), TempBaro = c(3.89, 5.1, 7.79, 8.89, 8.04, 8.01, 7.7, 
7.88, 7.25, 6.5), PressBaro = c(1053.98, 1054.5, 1054.53, 1054.71, 
1054.66, 1054.67, 1054.39, 1054.24, 1054.26, 1054.22), subPress = c(20.1099999999999, 
20.3699999999999, 20.27, 20.3099999999999, 20.29, 20.28, 20.1999999999998, 
20.21, 20.1900000000001, 20.3), subPressM = c(0.201099999999999, 
0.203699999999999, 0.2027, 0.203099999999999, 0.2029, 0.2028, 
0.201999999999998, 0.2021, 0.201900000000001, 0.203), subPressChange = c(0, 
0.00259999999999991, 0.00160000000000082, 0.00200000000000045, 
0.00180000000000063, 0.00170000000000073, 0.000899999999999179, 
0.00100000000000136, 0.000800000000001549, 0.00190000000000054
)), row.names = c(NA, 10L), class = "data.frame")

# 初始化relToDip向量
rel_to_dip <- numeric(nrow(df))
# 设置最后一行值为0.285
rel_to_dip[nrow(df)] <- 0.285

# 从倒数第二行向上循环计算
for(i in (nrow(df)-1):1) {
  rel_to_dip[i] <- rel_to_dip[i+1] + df$subPressChange[i]
}

# 将计算结果添加到DataFrame
df$relToDip <- rel_to_dip

方法二：dplyr 向量化实现 #

如果偏好tidyverse风格，可通过反向累加避免循环，效率更高：

library(dplyr)

df <- df %>%
  # 反转subPressChange后计算累加和，再反转回来得到反向累加结果
  mutate(relToDip = rev(cumsum(rev(subPressChange)) + 0.285 - last(subPressChange)))

结果说明 #

两种方法生成的relToDip列完全一致：最后一行固定为0.285，向上每一行的值等于下一行的relToDip加上当前行的subPressChange。

内容的提问来源于stack exchange，提问作者Melanie Baker