Update app.R

app.R

@@ -5,7 +5,7 @@ library(DT)
 library(ggplot2)
 library(shinyjs)
 
-# ── Load data ──────────────────────────────────────────────────────────────────
+# ── Load functions ─────────────────────────────────────────────────────────────
 load_data <- function(level) {
   if (level == "All") {
     dplyr::bind_rows(lapply(c("D1","D2","D3","JUCO","NAIA"), function(l) {
@@ -16,15 +16,23 @@ load_data <- function(level) {
   }
 }
 
-load_scores <- function(level) {
+load_pitcher_leaderboard <- function(level) {
   if (level == "All") {
     dplyr::bind_rows(lapply(c("D1","D2","D3","JUCO","NAIA"), function(l) {
-      arrow::read_parquet(paste0("/app/", l, "_scores.parquet"))
-    })) %>%
-      distinct(Pitcher, .keep_all = TRUE)
+      arrow::read_parquet(paste0("/app/", l, "_pitcher_leaderboard.parquet"))
+    }))
   } else {
-    arrow::read_parquet(paste0("/app/", level, "_scores.parquet")) %>%
-      distinct(Pitcher, .keep_all = TRUE)
+    arrow::read_parquet(paste0("/app/", level, "_pitcher_leaderboard.parquet"))
+  }
+}
+
+load_hitter_leaderboard <- function(level) {
+  if (level == "All") {
+    dplyr::bind_rows(lapply(c("D1","D2","D3","JUCO","NAIA"), function(l) {
+      arrow::read_parquet(paste0("/app/", l, "_hitter_leaderboard.parquet"))
+    }))
+  } else {
+    arrow::read_parquet(paste0("/app/", level, "_hitter_leaderboard.parquet"))
   }
 }
 
@@ -33,8 +41,7 @@ load_portal <- function() {
     df <- arrow::read_parquet("/app/transfer_portal.parquet")
     names(df) <- tolower(names(df))
     df %>%
-      rename(Pitcher = pitcher, Portal = transfer,
-             ERA = era, FIP = fip, WHIP = whip) %>%
+      rename(Pitcher=pitcher, Portal=transfer, ERA=era, FIP=fip, WHIP=whip) %>%
       distinct(Pitcher, .keep_all = TRUE)
   }, error = function(e) {
     message("Portal load failed: ", e$message)
@@ -87,201 +94,49 @@ clean_pitch_type <- function(pt) {
   )
 }
 
-# ── Pitcher stats ──────────────────────────────────────────────────────────────
-build_pitcher_stats <- function(df, min_pitches, scores, portal) {
-
-  team_lookup <- df %>%
-    select(Pitcher, PitcherTeam) %>%
-    distinct(Pitcher, .keep_all = TRUE)
-
-  df %>%
-    mutate(
-      IsStrike  = PitchCall %in% c("StrikeCalled","StrikeSwinging","FoulBall",
-                                   "FoulBallNotFieldable","FoulTip","InPlay"),
-      IsZone    = !is.na(PlateLocSide) & !is.na(PlateLocHeight) &
-        abs(PlateLocSide) <= 0.8303 & PlateLocHeight >= 1.5 & PlateLocHeight <= 3.3775,
-      IsWhiff   = PitchCall == "StrikeSwinging",
-      IsSwing   = PitchCall %in% c("StrikeSwinging","FoulBall",
-                                   "FoulBallNotFieldable","FoulTip","InPlay"),
-      IsChase   = IsSwing & !IsZone,
-      IsOutZone = !IsZone,
-      IsHardHit = !is.na(ExitSpeed) & ExitSpeed >= 95 &
-        !PitchCall %in% c("FoulBall","FoulBallNotFieldable","FoulBallFieldable","FoulTip"),
-      IsK       = !is.na(KorBB) & KorBB == "Strikeout",
-      IsBB      = !is.na(KorBB) & KorBB == "Walk",
-      IsPA      = (!is.na(KorBB) & KorBB %in% c("Strikeout","Walk")) |
-                  (!is.na(PitchCall) & PitchCall %in% c("InPlay","HitByPitch"))
-    ) %>%
-    group_by(Pitcher, Level) %>%
-    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) &
-                                 TaggedPitchType %in% c("Fastball","FourSeamFastBall",
-                                                        "Four-Seam","FourSeam",
-                                                        "Sinker","TwoSeamFastBall",
-                                                        "OneSeamFastball","Cutter"),
-                               RelSpeed, NA), na.rm = TRUE),
-      fb_velo_n   = sum(!is.na(RelSpeed) &
-                          TaggedPitchType %in% c("Fastball","FourSeamFastBall",
-                                                 "Four-Seam","FourSeam",
-                                                 "Sinker","TwoSeamFastBall",
-                                                 "OneSeamFastball","Cutter"),
-                        na.rm = TRUE),
-      .groups     = "drop"
-    ) %>%
-    left_join(team_lookup, by = "Pitcher") %>%
-    left_join(
-      scores %>% select(Pitcher, `Stuff+`, `Location+`) %>%
-        distinct(Pitcher, .keep_all = TRUE),
-      by = "Pitcher"
-    ) %>%
+# ── Build pitcher table from leaderboard ───────────────────────────────────────
+build_pitcher_table <- function(lb, min_pitches, portal_only) {
+  df <- lb %>%
     filter(Pitches >= min_pitches) %>%
-    left_join(portal %>% select(Pitcher, Portal, ERA, FIP, WHIP), by = "Pitcher") %>%
     mutate(
-      Portal      = ifelse(is.na(Portal), "No", as.character(Portal)),
-      ERA         = ifelse(Portal == "No", NA_real_, as.numeric(ERA)),
-      FIP         = ifelse(Portal == "No", NA_real_, as.numeric(FIP)),
-      WHIP        = ifelse(Portal == "No", NA_real_, as.numeric(WHIP)),
-      Name        = sapply(Pitcher, format_name),
-      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_,
-                      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)),
-      `Hard Hit%` = round(hh_n      / pmax(bip_n,     1) * 100, 1),
-      `Plot`      = paste0('<button class="plot-btn" data-pitcher="',
-                           gsub('"', '&quot;', Pitcher),
-                           '">&#128202; View</button>'),
-      `Twitter`   = paste0('<a href="https://www.google.com/search?q=',
-                           gsub(" ", "+", sapply(Pitcher, format_name)),
-                           '+baseball+twitter" target="_blank">&#x1F426; Search</a>')
-    ) %>%
-    filter(!is.na(Name), trimws(Name) != "", grepl(" ", trimws(Name))) %>%
-    select(Name, Level, Team, Portal, `Plot`, `Twitter`, Pitches, ERA, FIP, WHIP,
-           `FB Velo`, `Zone%`, `Strike%`, `K's`, `BB's`, `K/BB`,
-           `Whiff%`, `Chase%`, `Hard Hit%`, `Stuff+`, `Location+`) %>%
+      `Plot`    = paste0('<button class="plot-btn" data-pitcher="',
+                         gsub('"', '&quot;', Pitcher),
+                         '">&#128202; View</button>'),
+      `Twitter` = paste0('<a href="https://www.google.com/search?q=',
+                         gsub(" ", "+", Name),
+                         '+baseball+twitter" target="_blank">&#x1F426; Search</a>')
+    )
+  if (portal_only) df <- df %>% filter(Portal == "Yes")
+  df %>%
+    select(Name, Level, Team, Portal, `Plot`, `Twitter`, HomeState, Height, Weight,
+           Pitches, ERA, FIP, WHIP, `FB Velo`, `Zone%`, `Strike%`,
+           `K's`, `BB's`, `K/BB`, `Whiff%`, `Chase%`, `Hard Hit%`,
+           `Stuff+`, `Location+`) %>%
     arrange(desc(Pitches))
 }
 
-# ── Hitter stats ───────────────────────────────────────────────────────────────
-build_hitter_stats <- function(df, min_pa, hitter_portal) {
-
-  team_lookup <- df %>%
-    select(Batter, BatterTeam) %>%
-    distinct(Batter, .keep_all = TRUE)
-
-  df %>%
-    mutate(
-      IsSwing     = PitchCall %in% c("StrikeSwinging","FoulBall",
-                                     "FoulBallNotFieldable","FoulTip","InPlay"),
-      IsContact   = PitchCall %in% c("FoulBall","FoulBallNotFieldable","FoulTip","InPlay"),
-      IsWhiff     = PitchCall == "StrikeSwinging",
-      IsZone      = !is.na(PlateLocSide) & !is.na(PlateLocHeight) &
-        abs(PlateLocSide) <= 0.8303 & PlateLocHeight >= 1.5 & PlateLocHeight <= 3.3775,
-      IsChase     = IsSwing & !IsZone,
-      IsOutZone   = !IsZone,
-      IsZoneSwing = IsSwing & IsZone,
-      IsZoneWhiff = IsWhiff & IsZone,
-      IsHardHit   = !is.na(ExitSpeed) & ExitSpeed >= 95 &
-        !PitchCall %in% c("FoulBall","FoulBallNotFieldable","FoulBallFieldable","FoulTip"),
-      IsBIP       = !is.na(ExitSpeed) &
-        !PitchCall %in% c("FoulBall","FoulBallNotFieldable","FoulBallFieldable","FoulTip"),
-      IsGB        = IsBIP & TaggedHitType %in% c("Groundball","GroundBall"),
-      IsFB        = IsBIP & TaggedHitType == "FlyBall",
-      IsLD        = IsBIP & TaggedHitType == "LineDrive",
-      IsHit       = !is.na(PlayResult) & PlayResult %in% c("Single","Double","Triple","HomeRun"),
-      IsAB        = !is.na(PlayResult) & PlayResult %in% c("Single","Double","Triple","HomeRun",
-                                                            "Out","Error","FieldersChoice"),
-      IsPA        = (!is.na(KorBB) & KorBB %in% c("Strikeout","Walk")) |
-                    (!is.na(PitchCall) & PitchCall %in% c("InPlay","HitByPitch")),
-      IsHR        = !is.na(PlayResult) & PlayResult == "HomeRun",
-      IsBB        = !is.na(KorBB) & KorBB == "Walk",
-      IsK         = !is.na(KorBB) & KorBB == "Strikeout"
-    ) %>%
-    group_by(Batter, Level) %>%
-    summarise(
-      Pitches      = n(),
-      PA           = sum(IsPA,         na.rm = TRUE),
-      AB           = sum(IsAB,         na.rm = TRUE),
-      H            = sum(IsHit,        na.rm = TRUE),
-      HR           = sum(IsHR,         na.rm = TRUE),
-      BB           = sum(IsBB,         na.rm = TRUE),
-      K            = sum(IsK,          na.rm = TRUE),
-      hh_n         = sum(IsHardHit,    na.rm = TRUE),
-      bip_n        = sum(IsBIP,        na.rm = TRUE),
-      gb_n         = sum(IsGB,         na.rm = TRUE),
-      fb_n         = sum(IsFB,         na.rm = TRUE),
-      ld_n         = sum(IsLD,         na.rm = TRUE),
-      whiff_n      = sum(IsWhiff,      na.rm = TRUE),
-      swing_n      = sum(IsSwing,      na.rm = TRUE),
-      contact_n    = sum(IsContact,    na.rm = TRUE),
-      chase_n      = sum(IsChase,      na.rm = TRUE),
-      outzone_n    = sum(IsOutZone,    na.rm = TRUE),
-      zone_swing_n = sum(IsZoneSwing,  na.rm = TRUE),
-      zone_whiff_n = sum(IsZoneWhiff,  na.rm = TRUE),
-      ev_sum       = sum(ifelse(IsBIP, ExitSpeed, NA), na.rm = TRUE),
-      .groups      = "drop"
-    ) %>%
-    left_join(team_lookup, by = "Batter") %>%
+# ── Build hitter table from leaderboard ────────────────────────────────────────
+build_hitter_table <- function(lb, min_pa, portal_only, bat_side, hitter_pos) {
+  df <- lb %>%
     filter(PA >= min_pa) %>%
-    left_join(
-      hitter_portal %>% select(Batter, Portal, Position, BatSide, BA, OBP, SLG),
-      by = "Batter"
-    ) %>%
     mutate(
-      Portal      = ifelse(is.na(Portal), "No", as.character(Portal)),
-      BA          = ifelse(Portal == "No", NA_real_, as.numeric(BA)),
-      OBP         = ifelse(Portal == "No", NA_real_, as.numeric(OBP)),
-      SLG         = ifelse(Portal == "No", NA_real_, as.numeric(SLG)),
-      Name        = paste0('<button class="hitter-name-btn" data-batter="',
-                           gsub('"', '&quot;', Batter),
-                           '" style="background:none;border:none;color:#002147;',
-                           'font-weight:700;cursor:pointer;padding:0;font-size:13px;">',
-                           sapply(Batter, format_name), '</button>'),
-      Team        = BatterTeam,
-      AVG         = round(H   / pmax(AB, 1),                         3),
-      `BB%`       = round(BB  / pmax(PA, 1) * 100,                   1),
-      `K%`        = round(K   / pmax(PA, 1) * 100,                   1),
-      `Swing%`    = round(swing_n    / Pitches                 * 100, 1),
-      `Contact%`  = round(contact_n  / pmax(swing_n,        1) * 100, 1),
-      `IZ Whiff%` = round(zone_whiff_n / pmax(zone_swing_n,  1) * 100, 1),
-      `Chase%`    = round(chase_n    / pmax(outzone_n,      1) * 100, 1),
-      `Hard Hit%` = round(hh_n       / pmax(bip_n,          1) * 100, 1),
-      `Avg EV`    = round(ev_sum     / pmax(bip_n,          1),        1),
-      `GB%`       = round(gb_n       / pmax(bip_n,          1) * 100, 1),
-      `FB%`       = round(fb_n       / pmax(bip_n,          1) * 100, 1),
-      `LD%`       = round(ld_n       / pmax(bip_n,          1) * 100, 1),
-      `Twitter`   = paste0('<a href="https://www.google.com/search?q=',
-                           gsub(" ", "+", sapply(Batter, format_name)),
-                           '+baseball+twitter" target="_blank">&#x1F426; Search</a>')
-    ) %>%
-    filter(!is.na(Name), grepl("data-batter", Name)) %>%
-    select(Name, Level, Team, Portal, Position, BatSide, `Twitter`, PA,
-           AVG, OBP, SLG, `BB%`, `K%`, `Swing%`, `Contact%`, `IZ Whiff%`,
-           `Chase%`, `Hard Hit%`, `Avg EV`, `GB%`, `FB%`, `LD%`) %>%
+      Name = paste0('<button class="hitter-name-btn" data-batter="',
+                    gsub('"', '&quot;', Batter),
+                    '" style="background:none;border:none;color:#002147;',
+                    'font-weight:700;cursor:pointer;padding:0;font-size:13px;">',
+                    Name, '</button>'),
+      `Twitter` = paste0('<a href="https://www.google.com/search?q=',
+                         gsub(" ", "+", sapply(Batter, format_name)),
+                         '+baseball+twitter" target="_blank">&#x1F426; Search</a>')
+    )
+  if (portal_only)   df <- df %>% filter(Portal == "Yes")
+  if (bat_side != "All") df <- df %>% filter(BatSide == bat_side)
+  if (hitter_pos != "All") df <- df %>% filter(Position == hitter_pos)
+  df %>%
+    select(Name, Level, Team, Portal, Position, BatSide, `Twitter`,
+           HomeState, Height, Weight, PA,
+           AVG, OBP, SLG, BA, `BB%`, `K%`, `Swing%`, `Contact%`,
+           `IZ Whiff%`, `Chase%`, `Hard Hit%`, `Avg EV`, `GB%`, `FB%`, `LD%`) %>%
     arrange(desc(PA))
 }
 
@@ -422,7 +277,6 @@ ui <- fluidPage(
         Shiny.addCustomMessageHandler('update_label', function(msg) {
           $('#min_label').text(msg);
         });
-
       });
     "))
   ),
@@ -523,14 +377,14 @@ ui <- fluidPage(
 # ── Server ─────────────────────────────────────────────────────────────────────
 server <- function(input, output, session) {
 
+  # ── Load leaderboards once ──────────────────────────────────────────────────
   raw_data <- reactive({
     req(input$level)
     tryCatch({
       list(
-        df            = load_data(input$level),
-        scores        = load_scores(input$level),
-        portal        = load_portal(),
-        hitter_portal = load_hitter_portal()
+        pitcher_lb    = load_pitcher_leaderboard(input$level),
+        hitter_lb     = load_hitter_leaderboard(input$level),
+        raw_df        = load_data(input$level)
       )
     }, error = function(e) {
       showNotification(paste("Failed to load:", e$message), type = "error")
@@ -538,6 +392,7 @@ server <- function(input, output, session) {
     })
   })
 
+  # ── Position switch ─────────────────────────────────────────────────────────
   observeEvent(input$position, {
     if (input$position == "Pitcher") {
       session$sendCustomMessage("update_label", "Min Pitches")
@@ -552,44 +407,39 @@ server <- function(input, output, session) {
     }
   })
 
+  # ── Stats table ─────────────────────────────────────────────────────────────
   stats_table <- reactive({
     req(raw_data(), input$min_pitches)
     if (input$position == "Pitcher") {
-      tryCatch({
-        df <- build_pitcher_stats(
-          raw_data()$df,
+      tryCatch(
+        build_pitcher_table(
+          raw_data()$pitcher_lb,
           input$min_pitches,
-          raw_data()$scores,
-          raw_data()$portal
-        )
-        if (input$portal_only) df <- df %>% filter(Portal == "Yes")
-        df
-      }, error = function(e) {
-        message("build_pitcher_stats failed: ", e$message)
-        showNotification(paste("Stats error:", e$message), type = "error")
-        NULL
-      })
+          input$portal_only
+        ),
+        error = function(e) {
+          message("build_pitcher_table failed: ", e$message)
+          NULL
+        }
+      )
     } else {
-      tryCatch({
-        df <- build_hitter_stats(
-          raw_data()$df,
+      tryCatch(
+        build_hitter_table(
+          raw_data()$hitter_lb,
           input$min_pitches,
-          raw_data()$hitter_portal
-        )
-        if (input$portal_only) df <- df %>% filter(Portal == "Yes")
-        if (input$bat_side != "All")
-          df <- df %>% filter(BatSide == input$bat_side)
-        if (input$hitter_position != "All")
-          df <- df %>% filter(Position == input$hitter_position)
-        df
-      }, error = function(e) {
-        message("build_hitter_stats failed: ", e$message)
-        showNotification(paste("Stats error:", e$message), type = "error")
-        NULL
-      })
+          input$portal_only,
+          input$bat_side,
+          input$hitter_position
+        ),
+        error = function(e) {
+          message("build_hitter_table failed: ", e$message)
+          NULL
+        }
+      )
     }
   })
 
+  # ── Main UI ─────────────────────────────────────────────────────────────────
   output$main_ui <- renderUI({
     if (input$position == "Pitcher") {
       tagList(
@@ -610,6 +460,7 @@ server <- function(input, output, session) {
     }
   })
 
+  # ── Pitcher table ────────────────────────────────────────────────────────────
   output$pitcher_table <- renderDT({
     req(stats_table(), nrow(stats_table()) > 0)
     dt <- stats_table()
@@ -635,12 +486,12 @@ server <- function(input, output, session) {
         pageLength  = 25,
         lengthMenu  = c(25, 50, 100),
         scrollX     = TRUE,
-        order       = list(list(6, "desc")),
+        order       = list(list(9, "desc")),
         orderMulti  = FALSE,
         dom         = "lfrtip",
         columnDefs  = list(
           list(className = "dt-left",   targets = 0:1),
-          list(className = "dt-center", targets = 2:20),
+          list(className = "dt-center", targets = 2:23),
           list(orderable = FALSE,       targets = c(4, 5))
         ),
         initComplete = JS("function(settings, json) {
@@ -655,6 +506,7 @@ server <- function(input, output, session) {
       )
   })
 
+  # ── Hitter table ─────────────────────────────────────────────────────────────
   output$hitter_table <- renderDT({
     req(stats_table(), nrow(stats_table()) > 0)
     dt <- stats_table()
@@ -668,12 +520,12 @@ server <- function(input, output, session) {
         pageLength  = 25,
         lengthMenu  = c(25, 50, 100),
         scrollX     = TRUE,
-        order       = list(list(7, "desc")),
+        order       = list(list(10, "desc")),
         orderMulti  = FALSE,
         dom         = "lfrtip",
         columnDefs  = list(
           list(className = "dt-left",   targets = 0:1),
-          list(className = "dt-center", targets = 2:21),
+          list(className = "dt-center", targets = 2:25),
           list(orderable = FALSE,       targets = 6)
         ),
         initComplete = JS("function(settings, json) {
@@ -703,7 +555,7 @@ server <- function(input, output, session) {
 
   pitcher_plot_data <- reactive({
     req(current_pitcher(), raw_data())
-    raw_data()$df %>%
+    raw_data()$raw_df %>%
       filter(Pitcher == current_pitcher()) %>%
       mutate(PitchType = clean_pitch_type(TaggedPitchType)) %>%
       filter(!is.na(PitchType), !is.na(HorzBreak), !is.na(InducedVertBreak))
@@ -834,7 +686,7 @@ server <- function(input, output, session) {
 
   batter_data <- reactive({
     req(current_batter(), raw_data())
-    raw_data()$df %>%
+    raw_data()$raw_df %>%
       filter(Batter == current_batter()) %>%
       mutate(
         PitchGroup = case_when(