为何pivot_wider会将数值类型转为列表类型？求列均值解决方法

当pivot_wider后目标列变成列表类型，本质是每个分组下的RIGHT_GAZE_X有多个值，导致宽表单元格存储为列表。以下两种方法可解决：

方法1：将列表列转换为数值类型 #

场景1：每个列表仅含单个值 #

用map_dbl提取列表中的唯一数值，将列转为数值型：

library(tidyverse)

data <- read.csv("./b002003.csv", header = TRUE) %>%
  pivot_wider(
    names_from = TRIAL_INDEX,
    values_from = RIGHT_GAZE_X
  ) %>%
  # 针对第3、6列转换类型
  mutate(across(c(3, 6), ~map_dbl(.x, identity)))

# 正常计算行均值
data$rowmeans <- rowMeans(select(data, c(3, 6)))

场景2：每个列表含多个值 #

先对原数据按分组聚合取均值，再转宽表，避免生成列表列：

library(tidyverse)

data <- read.csv("./b002003.csv", header = TRUE) %>%
  # 替换...为你需要保留的所有非聚合列
  group_by(TRIAL_INDEX, ...) %>%
  summarise(RIGHT_GAZE_X = mean(RIGHT_GAZE_X), .groups = "drop") %>%
  pivot_wider(
    names_from = TRIAL_INDEX,
    values_from = RIGHT_GAZE_X
  )

data$rowmeans <- rowMeans(select(data, c(3, 6)))

方法2：直接对列表列计算均值 #

无需转换类型，用rowwise结合mean直接计算每行的列表均值：

library(tidyverse)

data <- read.csv("./b002003.csv", header = TRUE) %>%
  pivot_wider(
    names_from = TRIAL_INDEX,
    values_from = RIGHT_GAZE_X
  ) %>%
  rowwise() %>%
  mutate(rowmeans = mean(c(!!!select(., c(3, 6))))) %>%
  ungroup()

或者用pmap_dbl实现相同效果：

data$rowmeans <- pmap_dbl(select(data, c(3, 6)), ~mean(c(..1, ..2)))

内容的提问来源于stack exchange，提问作者111