Update app.R

app.R

@@ -3,20 +3,7 @@ library(dplyr)
 library(arrow)
 library(DT)
 
-# ── Load data ──────────────────────────────────────────────────────────────────
-load_data <- function(level) {
-  path <- paste0("/app/", gsub("[^A-Za-z0-9]","_",level), ".parquet")
-  arrow::read_parquet(path)
-}
-
-build_pitcher_stats <- function(df, min_pitches) {
-
-  format_name <- function(name) {
-    parts <- strsplit(name, ", ")[[1]]
-    if (length(parts) == 2) paste(trimws(parts[2]), trimws(parts[1])) else name
-  }
-
-  df %>%
+df %>%
     mutate(
       IsStrike  = PitchCall %in% c("StrikeCalled","StrikeSwinging","FoulBall",
                                    "FoulBallNotFieldable","FoulTip","InPlay"),
@@ -34,60 +21,61 @@ build_pitcher_stats <- function(df, min_pitches) {
       IsPA      = (!is.na(KorBB) & KorBB %in% c("Strikeout","Walk")) |
                   (!is.na(PitchCall) & PitchCall %in% c("InPlay","HitByPitch"))
     ) %>%
-    group_by(Pitcher, Level) %>%
+    group_by(Pitcher, Level, PitcherTeam) %>%
     summarise(
-      Pitches   = n(),
-      K         = sum(IsK,       na.rm=TRUE),
-      BB        = sum(IsBB,      na.rm=TRUE),
-      PA        = sum(IsPA,      na.rm=TRUE),
-      zone_n    = sum(IsZone,    na.rm=TRUE),
-      strike_n  = sum(IsStrike,  na.rm=TRUE),
-      whiff_n   = sum(IsWhiff,   na.rm=TRUE),
-      swing_n   = sum(IsSwing,   na.rm=TRUE),
-      chase_n   = sum(IsChase,   na.rm=TRUE),
-      outzone_n = sum(IsOutZone, na.rm=TRUE),
-      hh_n      = sum(IsHardHit, na.rm=TRUE),
-      bip_n     = sum(!is.na(ExitSpeed) &
-                    !PitchCall %in% c("FoulBall","FoulBallNotFieldable",
-                                      "FoulBallFieldable","FoulTip"), na.rm=TRUE),
-      fb_velo_sum = sum(ifelse(!is.na(RelSpeed) & 
+      Pitches     = n(),
+      K           = sum(IsK,       na.rm=TRUE),
+      BB          = sum(IsBB,      na.rm=TRUE),
+      PA          = sum(IsPA,      na.rm=TRUE),
+      zone_n      = sum(IsZone,    na.rm=TRUE),
+      strike_n    = sum(IsStrike,  na.rm=TRUE),
+      whiff_n     = sum(IsWhiff,   na.rm=TRUE),
+      swing_n     = sum(IsSwing,   na.rm=TRUE),
+      chase_n     = sum(IsChase,   na.rm=TRUE),
+      outzone_n   = sum(IsOutZone, na.rm=TRUE),
+      hh_n        = sum(IsHardHit, na.rm=TRUE),
+      bip_n       = sum(!is.na(ExitSpeed) &
+                      !PitchCall %in% c("FoulBall","FoulBallNotFieldable",
+                                        "FoulBallFieldable","FoulTip"), na.rm=TRUE),
+      fb_velo_sum = sum(ifelse(!is.na(RelSpeed) &
                                  TaggedPitchType %in% c("Fastball","FourSeamFastBall",
-                                                         "Four-Seam","FourSeam",
-                                                         "Sinker","TwoSeamFastBall",
-                                                         "OneSeamFastball","Cutter"),
+                                                        "Four-Seam","FourSeam",
+                                                        "Sinker","TwoSeamFastBall",
+                                                        "OneSeamFastball","Cutter"),
                                RelSpeed, NA), na.rm=TRUE),
-      fb_velo_n   = sum(!is.na(RelSpeed) & 
+      fb_velo_n   = sum(!is.na(RelSpeed) &
                           TaggedPitchType %in% c("Fastball","FourSeamFastBall",
-                                                  "Four-Seam","FourSeam",
-                                                  "Sinker","TwoSeamFastBall",
-                                                  "OneSeamFastball","Cutter"),
+                                                 "Four-Seam","FourSeam",
+                                                 "Sinker","TwoSeamFastBall",
+                                                 "OneSeamFastball","Cutter"),
                         na.rm=TRUE),
-      .groups   = "drop"
+      .groups     = "drop"
     ) %>%
+    ungroup() %>%
+    as.data.frame() %>%
     filter(Pitches >= min_pitches) %>%
     mutate(
       Name        = sapply(Pitcher, format_name),
-      Team = first(PitcherTeam),
+      Team        = PitcherTeam,
       `Zone%`     = round(zone_n    / Pitches            * 100, 1),
       `Strike%`   = round(strike_n  / Pitches            * 100, 1),
       `K's`       = K,
       `BB's`      = BB,
       `K/BB`      = case_when(
-                      K == 0  & BB == 0 ~ NA_real_,
-                      BB == 0           ~ NA_real_,   # undefined — infinite
-                      K == 0            ~ 0,
-                      TRUE              ~ round(K / BB, 2)
+                      K == 0 & BB == 0 ~ NA_real_,
+                      BB == 0          ~ NA_real_,
+                      K == 0           ~ 0,
+                      TRUE             ~ round(K / BB, 2)
                     ),
       `Whiff%`    = round(whiff_n   / pmax(swing_n,   1) * 100, 1),
       `Chase%`    = round(chase_n   / pmax(outzone_n, 1) * 100, 1),
-      `FB Velo`   = ifelse(fb_velo_n == 0, NA_real_, round(fb_velo_sum / fb_velo_n, 1)),
+      `FB Velo`   = ifelse(fb_velo_n == 0, NA_real_,
+                           round(fb_velo_sum / fb_velo_n, 1)),
       `Hard Hit%` = round(hh_n      / pmax(bip_n,     1) * 100, 1)
     ) %>%
-    # filter out blank or incomplete names
     filter(
       !is.na(Name),
       trimws(Name) != "",
-      # must have both first and last name (at least one space)
       grepl(" ", trimws(Name))
     ) %>%
     select(Name, Level, Team, Pitches, `FB Velo`, `Zone%`, `Strike%`,