Filtering By Times Appeared as an Observation

Question

I am doing a baseball-related project using utility players as the subject. My definition of a utility player for this is a player that either plays an infield and outfield position, plays 3 infield position, or plays catcher and an outfield/infield (not including 1B) position.

I would like to try to filter my dataset out by players who fit that definition, however, I don't know how to do this as all of the players are separated by their position. Here's an example of the dataset.

18    Addison Russell   CHC  2B  381.2  NA    NA  NA    0   NA    0   1  NA   1 110    88 0.800  23  NA  NA   NA  0.0
19    Addison Russell   CHC  SS  143.2  NA    NA  NA    0   NA    0   1  NA   1  37    24 0.649  17  NA  NA   NA -0.3
20 Adeiny Hechavarria - - -  2B  204.0  NA    NA  NA    1   NA    0  -2  NA  -1  43    33 0.767   7  NA  NA   NA  0.1
21 Adeiny Hechavarria - - -  3B   71.0  NA    NA  NA    0   NA    0   0  NA   0  12     9 0.750   0  NA  NA   NA  0.0
22 Adeiny Hechavarria - - -  SS  182.0  NA    NA  NA    0   NA    1   0  NA   1  53    39 0.736  10  NA  NA   NA  0.6

So for example of what I would like to do; in my definition of a utility player, Addison Russell doesn't count because he only plays 2 infield positions, however, Adeiny Hechavarria does as he plays 3. So how can I filter out the data set to have only players that match my definition? I've never really done anything like this, so I don't really even know where to start.

    > dput(head(utility_plyr_df))
structure(list(ï..Name = c("A.J. Pollock", "A.J. Pollock", "Aaron Altherr", 
"Aaron Hicks", "Aaron Judge", "Abraham Almonte"), Team = c("LAD", 
"LAD", "- - -", "NYY", "NYY", "ARI"), Pos = c("CF", "LF", "CF", 
"CF", "RF", "RF"), Inn = c(510, 145, 52, 499.1, 775.1, 54), rSZ = c(NA_integer_, 
NA_integer_, NA_integer_, NA_integer_, NA_integer_, NA_integer_
), rCERA = c(NA_integer_, NA_integer_, NA_integer_, NA_integer_, 
NA_integer_, NA_integer_), rSB = c(NA_integer_, NA_integer_, 
NA_integer_, NA_integer_, NA_integer_, NA_integer_), rGDP = c(NA_integer_, 
NA_integer_, NA_integer_, NA_integer_, NA_integer_, NA_integer_
), rARM = c(-1L, -1L, 0L, -1L, 3L, 1L), rGFP = c(0L, -1L, 0L, 
0L, 3L, 0L), rPM = c(-8L, 1L, 0L, 0L, 14L, 2L), rTS = c(NA, NA, 
NA, NA, NA, NA), DRS = c(-9L, -1L, 0L, -1L, 20L, 3L), BIZ = c(75L, 
17L, 14L, 93L, 112L, 10L), Plays = c(67L, 14L, 13L, 83L, 104L, 
10L), RZR = c(0.893, 0.824, 0.929, 0.892, 0.929, 1), OOZ = c(24L, 
11L, 5L, 32L, 73L, 5L), FSR = c(NA, NA, NA, NA, NA, NA), FRM = c(NA_real_, 
NA_real_, NA_real_, NA_real_, NA_real_, NA_real_), ARM = c(-1.9, 
-1.4, -0.2, -1.1, 4, 0.4), DPR = c(NA_real_, NA_real_, NA_real_, 
NA_real_, NA_real_, NA_real_), RngR = c(-4.6, 0.3, -0.2, 1.2, 
7.6, 1.1), ErrR = c(0.4, -0.1, 0.1, 0.5, 1.1, -0.2), UZR = c(-6.1, 
-1.1, -0.3, 0.6, 12.7, 1.4), UZR.150 = c(-20.2, -13.3, -4.9, 
0.8, 24.2, 53.2), Def = c(-5.2, -1.9, -0.2, 1.4, 8.7, 1.1), playerid = c(9256L, 
9256L, 11270L, 5297L, 15640L, 5486L), outfld = c(375, 375, 375, 
375, 375, 375), infld = c(487, 487, 487, 487, 487, 487), pos_1b = c(97, 
97, 97, 97, 97, 97), pos_cr = c(93, 93, 93, 93, 93, 93)), row.names = c(NA, 
6L), class = "data.frame")

New attempted code @Parfait

uzr.2019 <- within(uzr.2019, {
  pos_cr <- ave(ifelse(Pos == "C", 1, 0), 'ï..Name', FUN=sum)
  pos_1b <- ave(ifelse(Pos == "1B", 1, 0), 'ï..Name', FUN=sum)
  infld  <- ave(ifelse(Pos %in% c("C", "P", "1B", "2B", "3B", "SS") , 1, 0), 'ï..Name', FUN=sum)
  outfld <- ave(ifelse(Pos %in% c("LF", "CF", "RF"), 1, 0), 'ï..Name', FUN=sum)
})


# SUBSET DATA FRAME BY OR CONDITIONS


utility_plyr_df <- subset(uzr.2019, 
                          (infld >= 1 & oufld >= 1)   | # plays either an infield and outfield position
                            (infld == 3)                 | # plays 3 infield positions
                            (pos_cr == 1 & infld >= 1 & 
                               oufld >= 1 & pos_1b == 0)    # plays catcher and an outfield/infield (not including 1B) 
)

Do you count the number of infield position as the number of lines per name? — Cainã Max Couto-Silva, Oct 31 '20 at 02:58
Please add data using `dput` or something that we can copy and use. Read about [how to ask a good question](http://stackoverflow.com/help/how-to-ask) and [how to give a reproducible example](http://stackoverflow.com/questions/5963269). Are you looking for https://stackoverflow.com/questions/20204257/subset-data-frame-based-on-number-of-rows-per-group?noredirect=1&lq=1 ? — Ronak Shah, Oct 31 '20 at 03:29

Cainã Max Couto-Silva · Answer 1 · 2020-10-31T05:22:31.590

One logic to get it done is to first drop out the positions that don't match your criteria, then count the number of observations per player, dropping out those fewer than 3.

First, define your valid positions (if you don't have too much, otherwise you'll need a programmatic approach to get them):

valid_positions <- c("2B", "3B", "SS")

Supposing that df is your data.frame, V2 is your player's column, and V5 is your position's column:

Using data.table:

library(data.table)
dt <- as.data.table(df)
desired_output <- dt[V5 %in% valid_positions, if (.N > 2) .SD, by = V2]

Using dplyr:

library(dplyr)
desired_output <- df %>%
  filter(V5 %in% valid_positions) %>%
  count(V2) %>%
  filter(n > 2) %>%
  semi_join(df, ., by = "V2")

Desired output (I guess):

  V1     V2          V3 V4 V5  V6 V7 V8 V9 V10 V11 V12 V13 V14 V15 V16 V17   V18 V19 V20 V21 V22 V23
1 20 Adeiny Hechavarria  - 2B 204 NA NA NA   1  NA   0  -2  NA  -1  43  33 0.767   7  NA  NA  NA 0.1
2 21 Adeiny Hechavarria  - 3B  71 NA NA NA   0  NA   0   0  NA   0  12   9 0.750   0  NA  NA  NA 0.0
3 22 Adeiny Hechavarria  - SS 182 NA NA NA   0  NA   1   0  NA   1  53  39 0.736  10  NA  NA  NA 0.6

Okay, that helps a little bit. However, I am not just simply looking for guys who play 3 positions, an outfielder can play 3 positions, but they aren't a utility guy. How can I edit the code so it fits with the constraints I am looking for? — kaifranke, Oct 31 '20 at 04:45
I still don't know which is your "infield" and "outfielder" labels. So, I've edited the answer to add an option to define wanted values as a vector, using it to drop out the unwanted positions. Also, based @Ronak Shah comment, I kinda improve the code, though it's a very small modification. — Cainã Max Couto-Silva, Oct 31 '20 at 05:25

Parfait · Answer 2 · 2020-10-31T12:40:50.710

0

Consider creating flagged columns with ifelse, then run grouped counts by player with ave. Then subset by multiple conditions.

# ASSIGN NEW COLUMNS IN DATA FRAME
bb_df <- within(bb_df, {
            pos_cr <- ave(ifelse(Pos == "C", 1, 0), playerid, FUN=sum)
            pos_1b <- ave(ifelse(Pos == "1B", 1, 0), playerid, FUN=sum)
            infld  <- ave(ifelse(Pos %in% c("C", "P", "1B", "2B", "3B", "SS") , 1, 0), playerid, FUN=sum)
            outfld <- ave(ifelse(Pos %in% c("LF", "CF", "RF"), 1, 0), player, FUN=sum)
         })

# SUBSET DATA FRAME BY OR CONDITIONS
utility_plyr_df <- subset(bb_df, 
                           (infld >= 1 & outfld >= 1)   | # plays either an infield and outfield position
                           (infld == 3)                 | # plays 3 infield positions
                           (pos_cr == 1 & infld >= 1 & 
                            outfld >= 1 & pos_1b == 0)    # plays catcher and an outfield/infield (not including 1B) 
                   )

edited Oct 31 '20 at 12:40

answered Oct 31 '20 at 04:34

Parfait

104,375
17
94
125

I ran this code, however, it is only adding up the amount of players that play each position for the pos_cr, pos_1b, infld, and outfld variables instead of just adding it up for each player individually. – kaifranke Oct 31 '20 at 05:54
I added the dput to be able to see what happened with the code. – kaifranke Oct 31 '20 at 05:57
Not quite sure how you adjusted code to fit actual data. `ave` runs calculations for *each* player not all if you used appropriate identifier like `playerid`. If other groupings like team, season, or year is in data, add those to `ave` before `FUN` arg. Testing showed some syntax change needs. See edit. And your `dput` sample is not large enough to find any utility players. So I adjust infield condition of 3 to outfield of 2 to demo: https://rextester.com/NWKJWH18763 – Parfait Oct 31 '20 at 12:44
And what's the issue? My demo shows code should work. Can you post a fuller dataset, maybe 50 rows in a pastebin link? – Parfait Oct 31 '20 at 21:21

Filtering By Times Appeared as an Observation

2 Answers2