RWEP/SD/20240326_3_datatransform/index.qmd

328 lines
4.1 KiB
Plaintext
Raw Blame History

This file contains ambiguous Unicode characters

This file contains Unicode characters that might be confused with other characters. If you think that this is intentional, you can safely ignore this warning. Use the Escape button to reveal them.

---
title: "Data Transform"
subtitle: 《区域水环境污染数据分析实践》<br>Data analysis practice of regional water environment pollution
author: 苏命、王为东<br>中国科学院大学资源与环境学院<br>中国科学院生态环境研究中心
date: today
lang: zh
format:
revealjs:
theme: dark
slide-number: true
chalkboard:
buttons: true
preview-links: auto
lang: zh
toc: true
toc-depth: 1
toc-title: 大纲
logo: ./_extensions/inst/img/ucaslogo.png
css: ./_extensions/inst/css/revealjs.css
pointer:
key: "p"
color: "#32cd32"
pointerSize: 18
revealjs-plugins:
- pointer
filters:
- d2
---
```{r}
#| echo: false
knitr::opts_chunk$set(echo = TRUE)
source("../../coding/_common.R")
library(nycflights13)
library(tidyverse)
```
## `tidyverse`风格数据分析总体流程
![](../../image/data-science/transform.png)
## [dplyr cheatsheet](../../image/cheatsheet/data-transformation.pdf)
```{r}
#| echo: false
dwfun::ggsavep("../../image/cheatsheet/data-transformation.svg", loadit = TRUE)
```
## 查看数据
```{r}
flights
```
## 选择列
```{r}
#| results: false
flights |>
select(year, month, day)
```
## 选择列
```{r}
#| results: false
flights |>
select(year:day)
```
## 选择列
```{r}
flights |>
select(3:5)
```
## 选择列
```{r}
flights |>
select(!year:day)
```
## 选择列
```{r}
flights |>
select(-(year:day))
```
## 选择列
```{r}
flights |>
select(where(is.character))
```
## 选择列
```{r}
flights |>
select(!where(is.character)) |>
select(contains("_"))
```
## 选择列
```{r}
flights |>
select(tail_num = tailnum)
```
## 选择列
```{r}
flights |>
select(air_time, everything())
```
## 重命名
```{r}
flights |>
rename(tail_num = tailnum)
```
## 重命名
```{r}
flights |>
rename(年份 = 1) |>
rename(月份 = 2)
```
## 重命名
```{r}
flights |> select(1:4) |> head(n = 3)
# 重命名
flights |> select(1:4) |> head(n = 3) |>
rename_all(~c("c1", "c2", "c3", "c4"))
```
## 重命名
```{r}
flights |> select(1:4) |> head(n = 3)
# 重命名
flights |> select(1:4) |> head(n = 3) |>
rename_all(toupper)
```
## 重命名
```{r}
flights |> select(1:4) |> head(n = 3)
# 重命名
flights |> select(1:4) |> head(n = 3) |>
rename_all(~paste0(toupper(.), "_NEW"))
```
## 练习
将含有下划线的列名中的下划线去掉。
```{r}
flights |> select(1:4) |> head(n = 3)
```
## 练习
将`airqualitydf`中列名的单位信息去除前5列
```{r}
airqualitydf <- readxl::read_xlsx("../../data/airquality.xlsx", sheet = 2)
airqualitydf |> select(1:5)
```
## `filter`
```{r}
flights |>
filter(dep_delay > 120)
```
## filter 练习
Flights that departed on January 1.
```{r}
#| echo: false
flights |>
filter(month == 1 & day == 1)
```
## filter 练习
Select flights that departed in January or February
```{r}
#| echo: false
flights |>
filter(month %in% c(1, 2))
```
## filter 练习
```{r}
jan1 <- flights |>
filter(month == 1 & day == 1)
```
## filter
```{r}
#| error: true
#| eval: false
flights |>
filter(month = 1)
```
## filter
```{r}
flights |>
filter(month == 1 | 2)
```
## 排序
```{r}
flights |>
arrange(year, month, day, dep_time)
```
## 排序
```{r}
flights |>
arrange(desc(dep_delay))
```
## slice
```{r}
flights |> head(n = 5)
flights |> slice(1:5)
```
## slice
```{r}
flights |>
slice_max(dep_delay, n = 5)
```
## slice
```{r}
flights |>
slice_min(dep_delay, prop = 0.005)
```
## 排序练习
根据`origin`、`dest`、`air_time`倒序排序。
```{r}
#| echo: false
flights |>
arrange(origin, dest, desc(air_time)) |>
select(origin, dest, air_time, everything())
```
## 去重
```{r}
# Remove duplicate rows, if any
flights |>
distinct()
```
## 去重
```{r}
# Find all unique origin and destination pairs
flights |>
distinct(origin, dest)
```
## 去重
```{r}
flights |>
distinct(origin, dest, .keep_all = TRUE)
```
## 欢迎讨论!{.center}
`r rmdify::slideend(wechat = FALSE, type = "public", tel = FALSE, thislink = "https://drwater.rcees.ac.cn/course/public/RWEP/@PUB/SD/")`