在dplyr中提供的动词语法与其他操作进行连接，可以实现更加复杂的数据操作

group操作

  1. # group操作
  2. by_tailnum <- group_by(flights, tailnum)
  4. # group操作
  5. delay <- summarise(by_tailnum,
  6.   count = n(),
  7.   dist = mean(distance, na.rm = TRUE),
  8.   delay = mean(arr_delay, na.rm = TRUE))
  1. delay
  1. ## # A tibble: 4,044 x 4
  2. ##    tailnum count  dist  delay
  3. ##    <chr>   <int> <dbl>  <dbl>
  4. ##  1 D942DN      4  854. 31.5  
  5. ##  2 N0EGMQ    371  676.  9.98 
  6. ##  3 N10156    153  758. 12.7  
  7. ##  4 N102UW     48  536.  2.94 
  8. ##  5 N103US     46  535. -6.93 
  9. ##  6 N104UW     47  535.  1.80 
  10. ##  7 N10575    289  520. 20.7  
  11. ##  8 N105UW     45  525. -0.267
  12. ##  9 N107US     41  529. -5.73 
  13. ## 10 N108UW     60  534. -1.25 
  14. ## # ... with 4,034 more rows

在summarise()函数中可以与基础的R函数(如：min())进行联合操作，同时dplyr同样提供了一系列的快捷操作：

  • n(): 观测值的个数
  • n_distinct(x): unique的观测值数目
  • first(x),last(x), nth(x, n): 第一个、最后一个、第n个观测值
  1. destinations <- group_by(flights, dest)
  2. summarise(destinations,
  3.   planes = n_distinct(tailnum),
  4.   flights = n()
  5. )
  1. ## # A tibble: 105 x 3
  2. ##    dest  planes flights
  3. ##    <chr>  <int>   <int>
  4. ##  1 ABQ      108     254
  5. ##  2 ACK       58     265
  6. ##  3 ALB      172     439
  7. ##  4 ANC        6       8
  8. ##  5 ATL     1180   17215
  9. ##  6 AUS      993    2439
  10. ##  7 AVL      159     275
  11. ##  8 BDL      186     443
  12. ##  9 BGR       46     375
  13. ## 10 BHM       45     297
  14. ## # ... with 95 more rows

当依据多个变量进行分组时，可以根据summarise()层层递进，得出结果

  1. daily <- group_by(flights, year, month, day)
  3. per_day   <- summarise(daily, flights = n())
  1. per_month <- summarise(per_day, flights = sum(flights))
  1. per_year  <- summarise(per_month, flights = sum(flights))

select 操作

select函数可以接受数字和具体的列名，但需要注意的是，当有与列名相同的其他赋值时，不能进行操作

  1. select(flights, year)
  1. ## # A tibble: 336,776 x 1
  2. ##     year
  3. ##    <int>
  4. ##  1  2013
  5. ##  2  2013
  6. ##  3  2013
  7. ##  4  2013
  8. ##  5  2013
  9. ##  6  2013
  10. ##  7  2013
  11. ##  8  2013
  12. ##  9  2013
  13. ## 10  2013
  14. ## # ... with 336,766 more rows
  1. select(flights, 1)
  1. ## # A tibble: 336,776 x 1
  2. ##     year
  3. ##    <int>
  4. ##  1  2013
  5. ##  2  2013
  6. ##  3  2013
  7. ##  4  2013
  8. ##  5  2013
  9. ##  6  2013
  10. ##  7  2013
  11. ##  8  2013
  12. ##  9  2013
  13. ## 10  2013
  14. ## # ... with 336,766 more rows
  1. year <- 5
  2. select(flights, year)
  1. ## # A tibble: 336,776 x 1
  2. ##     year
  3. ##    <int>
  4. ##  1  2013
  5. ##  2  2013
  6. ##  3  2013
  7. ##  4  2013
  8. ##  5  2013
  9. ##  6  2013
  10. ##  7  2013
  11. ##  8  2013
  12. ##  9  2013
  13. ## 10  2013
  14. ## # ... with 336,766 more rows

Mutating 操作

  1. df <- select(flights, year:dep_time)
  3. mutate(df, "year", 2)
  1. ## # A tibble: 336,776 x 6
  2. ##     year month   day dep_time `"year"`   `2`
  3. ##    <int> <int> <int>    <int> <chr>    <dbl>
  4. ##  1  2013     1     1      517 year         2
  5. ##  2  2013     1     1      533 year         2
  6. ##  3  2013     1     1      542 year         2
  7. ##  4  2013     1     1      544 year         2
  8. ##  5  2013     1     1      554 year         2
  9. ##  6  2013     1     1      554 year         2
  10. ##  7  2013     1     1      555 year         2
  11. ##  8  2013     1     1      557 year         2
  12. ##  9  2013     1     1      557 year         2
  13. ## 10  2013     1     1      558 year         2
  14. ## # ... with 336,766 more rows
  1. mutate(df, year + 10)
  1. ## # A tibble: 336,776 x 5
  2. ##     year month   day dep_time `year + 10`
  3. ##    <int> <int> <int>    <int>       <dbl>
  4. ##  1  2013     1     1      517        2023
  5. ##  2  2013     1     1      533        2023
  6. ##  3  2013     1     1      542        2023
  7. ##  4  2013     1     1      544        2023
  8. ##  5  2013     1     1      554        2023
  9. ##  6  2013     1     1      554        2023
  10. ##  7  2013     1     1      555        2023
  11. ##  8  2013     1     1      557        2023
  12. ##  9  2013     1     1      557        2023
  13. ## 10  2013     1     1      558        2023
  14. ## # ... with 336,766 more rows
  1. var <- seq(1, nrow(df))
  2. mutate(df, new = var)
  1. ## # A tibble: 336,776 x 5
  2. ##     year month   day dep_time   new
  3. ##    <int> <int> <int>    <int> <int>
  4. ##  1  2013     1     1      517     1
  5. ##  2  2013     1     1      533     2
  6. ##  3  2013     1     1      542     3
  7. ##  4  2013     1     1      544     4
  8. ##  5  2013     1     1      554     5
  9. ##  6  2013     1     1      554     6
  10. ##  7  2013     1     1      555     7
  11. ##  8  2013     1     1      557     8
  12. ##  9  2013     1     1      557     9
  13. ## 10  2013     1     1      558    10
  14. ## # ... with 336,766 more rows

group_by()的操作同样可以修改和创建变量，不同的时该变量是group之后的

  1. group_by(df, month)
  1. ## # A tibble: 336,776 x 4
  2. ## # Groups:   month [12]
  3. ##     year month   day dep_time
  4. ##    <int> <int> <int>    <int>
  5. ##  1  2013     1     1      517
  6. ##  2  2013     1     1      533
  7. ##  3  2013     1     1      542
  8. ##  4  2013     1     1      544
  9. ##  5  2013     1     1      554
  10. ##  6  2013     1     1      554
  11. ##  7  2013     1     1      555
  12. ##  8  2013     1     1      557
  13. ##  9  2013     1     1      557
  14. ## 10  2013     1     1      558
  15. ## # ... with 336,766 more rows
  1. group_by(df, month = as.factor(month))
  1. ## # A tibble: 336,776 x 4
  2. ## # Groups:   month [12]
  3. ##     year month   day dep_time
  4. ##    <int> <fct> <int>    <int>
  5. ##  1  2013 1         1      517
  6. ##  2  2013 1         1      533
  7. ##  3  2013 1         1      542
  8. ##  4  2013 1         1      544
  9. ##  5  2013 1         1      554
  10. ##  6  2013 1         1      554
  11. ##  7  2013 1         1      555
  12. ##  8  2013 1         1      557
  13. ##  9  2013 1         1      557
  14. ## 10  2013 1         1      558
  15. ## # ... with 336,766 more rows
  1. group_by(df, day_binned = cut(day, 3))
  1. ## # A tibble: 336,776 x 5
  2. ## # Groups:   day_binned [3]
  3. ##     year month   day dep_time day_binned
  4. ##    <int> <int> <int>    <int> <fct>     
  5. ##  1  2013     1     1      517 (0.97,11] 
  6. ##  2  2013     1     1      533 (0.97,11] 
  7. ##  3  2013     1     1      542 (0.97,11] 
  8. ##  4  2013     1     1      544 (0.97,11] 
  9. ##  5  2013     1     1      554 (0.97,11] 
  10. ##  6  2013     1     1      554 (0.97,11] 
  11. ##  7  2013     1     1      555 (0.97,11] 
  12. ##  8  2013     1     1      557 (0.97,11] 
  13. ##  9  2013     1     1      557 (0.97,11] 
  14. ## 10  2013     1     1      558 (0.97,11] 
  15. ## # ... with 336,766 more rows

pipe 操作

当进行多重的动词语法操作时，可以把一系列的操作进行融合，不过因此会损害代码的可读性

  1. # 鎸笺父鏄笺工鎴笺父鏄笺付鎸笺功鏀笺赴同鎴笺傅牟鎼笺腹鎼笺阜鏄笺阜 
  3. # method 1
  4. a1 <- group_by(flights, year, month, day)
  5. a2 <- select(a1, arr_delay, dep_delay)
  1. ## Adding missing grouping variables: `year`, `month`, `day`
  1. a3 <- summarise(a2,
  2.   arr = mean(arr_delay, na.rm = TRUE),
  3.   dep = mean(dep_delay, na.rm = TRUE))
  1. ## `summarise()` regrouping output by 'year', 'month' (override with `.groups` argument)
  1. a4 <- filter(a3, arr > 30 | dep > 30)
  2. a4
  1. ## # A tibble: 49 x 5
  2. ## # Groups:   year, month [11]
  3. ##     year month   day   arr   dep
  4. ##    <int> <int> <int> <dbl> <dbl>
  5. ##  1  2013     1    16  34.2  24.6
  6. ##  2  2013     1    31  32.6  28.7
  7. ##  3  2013     2    11  36.3  39.1
  8. ##  4  2013     2    27  31.3  37.8
  9. ##  5  2013     3     8  85.9  83.5
  10. ##  6  2013     3    18  41.3  30.1
  11. ##  7  2013     4    10  38.4  33.0
  12. ##  8  2013     4    12  36.0  34.8
  13. ##  9  2013     4    18  36.0  34.9
  14. ## 10  2013     4    19  47.9  46.1
  15. ## # ... with 39 more rows
  1. # method 2
  2. filter(
  3.   summarise(
  4.     select(
  5.       group_by(flights, year, month, day),
  6.       arr_delay, dep_delay
  7.     ),
  8.     arr = mean(arr_delay, na.rm = TRUE),
  9.     dep = mean(dep_delay, na.rm = TRUE)
  10.   ),
  11.   arr > 30 | dep > 30
  12. )
  1. ## Adding missing grouping variables: `year`, `month`, `day`
  1. ## `summarise()` regrouping output by 'year', 'month' (override with `.groups` argument)
  1. ## # A tibble: 49 x 5
  2. ## # Groups:   year, month [11]
  3. ##     year month   day   arr   dep
  4. ##    <int> <int> <int> <dbl> <dbl>
  5. ##  1  2013     1    16  34.2  24.6
  6. ##  2  2013     1    31  32.6  28.7
  7. ##  3  2013     2    11  36.3  39.1
  8. ##  4  2013     2    27  31.3  37.8
  9. ##  5  2013     3     8  85.9  83.5
  10. ##  6  2013     3    18  41.3  30.1
  11. ##  7  2013     4    10  38.4  33.0
  12. ##  8  2013     4    12  36.0  34.8
  13. ##  9  2013     4    18  36.0  34.9
  14. ## 10  2013     4    19  47.9  46.1
  15. ## # ... with 39 more rows
  1. # method 3
  3. flights %>%
  4.   group_by(year, month, day) %>%
  5.   select(arr_delay, dep_delay) %>%
  6.   summarise(
  7.     arr = mean(arr_delay, na.rm = TRUE),
  8.     dep = mean(dep_delay, na.rm = TRUE)
  9.   ) %>%
  10.   filter(arr > 30 | dep > 30)
  1. ## Adding missing grouping variables: `year`, `month`, `day`
  2. ## `summarise()` regrouping output by 'year', 'month' (override with `.groups` argument)
  1. ## # A tibble: 49 x 5
  2. ## # Groups:   year, month [11]
  3. ##     year month   day   arr   dep
  4. ##    <int> <int> <int> <dbl> <dbl>
  5. ##  1  2013     1    16  34.2  24.6
  6. ##  2  2013     1    31  32.6  28.7
  7. ##  3  2013     2    11  36.3  39.1
  8. ##  4  2013     2    27  31.3  37.8
  9. ##  5  2013     3     8  85.9  83.5
  10. ##  6  2013     3    18  41.3  30.1
  11. ##  7  2013     4    10  38.4  33.0
  12. ##  8  2013     4    12  36.0  34.8
  13. ##  9  2013     4    18  36.0  34.9
  14. ## 10  2013     4    19  47.9  46.1
  15. ## # ... with 39 more rows