Hey there! No worries at all about being new or not knowing all the jargon—we’ve all been stuck on these tidy data hurdles when starting out 😊

You’re totally right that tidyr is the go-to package for reshaping data, and spread() does handle long-to-wide conversions. The catch is that spread() is designed to work with one value column at a time, which is probably why you hit a roadblock when trying to merge multiple columns. Let’s break this down with examples, and also cover a more flexible (and modern) alternative since spread() is now deprecated in newer versions of tidyr.

示例场景 #

First, let’s use a sample long-format dataset that matches your need to keep multiple value columns (like a numeric value and its unit):

library(tidyr)
library(dplyr) # 用来辅助数据整理

# 模拟长格式数据
long_df <- tibble(
  person_id = rep(1:3, each = 2),
  measure = c("height", "weight", "height", "weight", "height", "weight"),
  value = c(170, 70, 165, 65, 180, 80),
  unit = c("cm", "kg", "cm", "kg", "cm", "kg")
)

方法1：用spread()处理多列 #

Since spread() only works with one value column, we’ll split the task into steps:

1. Spread each value column separately
2. Merge the resulting wide data frames back together

# 第一步：spread数值列
wide_values <- long_df %>%
  select(person_id, measure, value) %>%
  spread(key = measure, value = value)

# 第二步：spread单位列
wide_units <- long_df %>%
  select(person_id, measure, unit) %>%
  spread(key = measure, value = unit)

# 第三步：合并两个宽数据框，用后缀区分列名
final_wide <- wide_values %>%
  left_join(wide_units, by = "person_id", suffix = c("_value", "_unit"))

This will give you a data frame with columns: person_id, height_value, weight_value, height_unit, weight_unit.

方法2：用pivot_wider()（推荐！） #

pivot_wider() is the modern replacement for spread() and handles multiple value columns in one step—way more efficient! Here’s how to use it:

final_wide <- long_df %>%
  pivot_wider(
    id_cols = person_id, # 用来分组的唯一标识列
    names_from = measure, # 要转成宽列的分类列
    values_from = c(value, unit), # 要保留的多列值
    names_glue = "{measure}_{.value}" # 自定义列名格式，可选
  )

The names_glue argument lets you format column names exactly how you want—this example will create columns like height_value, weight_unit, etc. If you prefer value_height instead, just swap the order in names_glue: "{.value}_{measure}".

为什么你之前用spread()失败了？ #

Chances are you tried passing multiple columns to the value argument of spread(), which it doesn’t support. spread() expects one key column and one value column—so when you have multiple value columns, you need to either split the task (method 1) or use the more flexible pivot_wider() (method 2).

内容的提问来源于stack exchange，提问作者B.P.