Question

我正在对大型data.frame进行数据分析。有没有一种方法可以根据过滤后的标准保留完整的data.frame来进行排名？该代码的工作原理与使用dplyr filter结果相似，但是在data.frame

之后保留了整个mutate的内容

library(tidyverse)

# Data
df <- read.table(sep="\t", text="
namePlayer    groupPosition   minutesTotals   fgmTotals   fgaTotals   
fg3mTotals  fg3aTotals  fg2mTotals  fg2aTotals  ftmTotals   ftaTotals   
orbTotals   drbTotals   trbTotals   astTotals   stlTotals   blkTotals   
tovTotals   pfTotals    ptsTotals
             Anthony Davis  C   1267    353 698  34 105 319 593 236 294 114 340 454 151 58  90   71  86  976
             Bradley Beal   G   1392    336 714  89 262 247 452 137 174  36 148 184 180 47  32   98 115  898
             Damian Lillard G   1347    335 741 117 297 218 444 229 256  30 151 181 219 35  19  104  69 1016
             Giannis Antetokounmpo  F   1146    335 572  12  79 323 493 212 305  83 346 429 207 45  54  145 116  894
             James Harden   G   1261    331 752 162 416 169 336 318 374  26 175 201 291 70  19  189 116 1142
             Joel Embiid    C   1255    330 680  42 142 288 538 295 368  89 413 502 129 21  70  128 125  997
             Kemba Walker   G   1276    324 734 122 335 202 399 171 208  16 142 158 223 48  18   89  61  941
             Kevin Durant   F   1399    383 760  70 192 313 568 274 300  19 274 293 237 32  39  133  74 1110
             LeBron James   F   1178    340 656  68 191 272 465 180 264  32 251 283 243 44  24  116  54  928
             Paul George    F   1271    332 734 119 315 213 419 179 215  56 235 291 146 82  22   99 106  962", header=TRUE, stringsAsFactors=FALSE)

df_calc <- df %>%
  # Overall Rank
  mutate(o_rank = rank(desc(ptsTotals))) %>%
  # Rank by Position
  group_by(groupPosition) %>%
  mutate(position_rank = rank(desc(ptsTotals))) %>%
  ungroup() %>%
  # Conditional Rank
  mutate(custom_rank = ifelse(groupPosition %in% c("G", "F") & position_rank > 3 |
                          groupPosition =="C" & position_rank > 3, rank(desc(ptsTotals)), NA ))


df_calc_correct <-  df %>%
  # Overall Rank
  mutate(o_rank = rank(desc(ptsTotals))) %>%
  # Rank by Position
  group_by(groupPosition) %>%
  mutate(position_rank = rank(desc(ptsTotals))) %>%
  ungroup() %>%
  # Conditional Rank
  filter(groupPosition %in% c("G", "F") & position_rank > 3 |
                            groupPosition =="C" & position_rank > 3) %>% 
  mutate(custom_rank = rank(desc(ptsTotals)))

df_calc
#> # A tibble: 10 x 23
#>    namePlayer groupPosition minutesTotals fgmTotals fgaTotals fg3mTotals
#>    <chr>      <chr>                 <int>     <int>     <int>      <int>
#>  1 "        ~ C                      1267       353       698         34
#>  2 "        ~ G                      1392       336       714         89
#>  3 "        ~ G                      1347       335       741        117
#>  4 "        ~ F                      1146       335       572         12
#>  5 "        ~ G                      1261       331       752        162
#>  6 "        ~ C                      1255       330       680         42
#>  7 "        ~ G                      1276       324       734        122
#>  8 "        ~ F                      1399       383       760         70
#>  9 "        ~ F                      1178       340       656         68
#> 10 "        ~ F                      1271       332       734        119
#> # ... with 17 more variables: fg3aTotals <int>, fg2mTotals <int>,
#> #   fg2aTotals <int>, ftmTotals <int>, ftaTotals <int>, orbTotals <int>,
#> #   drbTotals <int>, trbTotals <int>, astTotals <int>, stlTotals <int>,
#> #   blkTotals <int>, tovTotals <int>, pfTotals <int>, ptsTotals <int>,
#> #   o_rank <dbl>, position_rank <dbl>, custom_rank <dbl>

df_calc_correct
#> # A tibble: 2 x 23
#>   namePlayer groupPosition minutesTotals fgmTotals fgaTotals fg3mTotals
#>   <chr>      <chr>                 <int>     <int>     <int>      <int>
#> 1 "        ~ G                      1392       336       714         89
#> 2 "        ~ F                      1146       335       572         12
#> # ... with 17 more variables: fg3aTotals <int>, fg2mTotals <int>,
#> #   fg2aTotals <int>, ftmTotals <int>, ftaTotals <int>, orbTotals <int>,
#> #   drbTotals <int>, trbTotals <int>, astTotals <int>, stlTotals <int>,
#> #   blkTotals <int>, tovTotals <int>, pfTotals <int>, ptsTotals <int>,
#> #   o_rank <dbl>, position_rank <dbl>, custom_rank <dbl>

由reprex软件包（v0.2.1）于2019-01-04创建

Answer 1

我无法正确读取您的df，但是这种常规方法应该可以

set.seed(1)
df <- data.frame(a = 1:10, b = sample(1:10))

df %>% 
  mutate(custom_rank = {
    filt <- a %in% 3:5
    replace(rep(NA, n()), which(filt), rank(desc(b[filt])))})

#     a  b custom_rank
# 1   1  3          NA
# 2   2  4          NA
# 3   3  5           2
# 4   4  7           1
# 5   5  2           3
# 6   6  8          NA
# 7   7  9          NA
# 8   8  6          NA
# 9   9 10          NA
# 10 10  1          NA

Answer 2

无法加载表格，但是如果为正ifelse条件过滤ptsTotals向量，它应该可以工作：

df %>%
  # Overall Rank
  mutate(o_rank = rank(desc(ptsTotals))) %>%
  # Rank by Position
  group_by(groupPosition) %>%
  mutate(position_rank = rank(desc(ptsTotals))) %>%
  ungroup() %>%
  # Conditional Rank
  mutate(custom_rank = ifelse(groupPosition %in% c("G", "F") & position_rank > 3 |
                                groupPosition =="C" & position_rank > 3,
                              rank(desc(ptsTotals[groupPosition %in% c("G", "F") & position_rank > 3 |
                                                    groupPosition =="C" & position_rank > 3])), NA ))

有没有办法做一个保留完整数据帧的过滤等级？

2 个答案: