Update app.R

app.R

@@ -4,8 +4,6 @@ library(arrow)
 library(DT)
 library(ggplot2)
 library(shinyjs)
-library(tidymodels)
-library(xgboost)
 
 # ── Load data ──────────────────────────────────────────────────────────────────
 load_data <- function(level) {
@@ -13,11 +11,10 @@ load_data <- function(level) {
   arrow::read_parquet(path)
 }
 
-# ── Models ─────────────────────────────────────────────────────────────────────
-model                <- readRDS("/app/Stuff+2.rds")
-league_stats         <- readRDS("/app/NEW_LeagueStats2.rds")
-xgb_fit              <- readRDS("/app/location_plus_model.rds")
-league_stats_pitcher <- readRDS("/app/location_plus_league_stats_pitcher.rds")
+load_scores <- function(level) {
+  path <- paste0("/app/", level, "_scores.parquet")
+  arrow::read_parquet(path)
+}
 
 # ── Pitch colors ───────────────────────────────────────────────────────────────
 pitch_colors <- c(
@@ -47,113 +44,8 @@ clean_pitch_type <- function(pt) {
   )
 }
 
-map_pitch_type <- function(pt) {
-  case_when(
-    pt %in% c("Fastball","FourSeamFastBall","Four-Seam","FourSeam") ~ "Fastball",
-    pt %in% c("Sinker","TwoSeamFastBall","OneSeamFastball")         ~ "Sinker",
-    pt == "Cutter"                                                   ~ "Cutter",
-    pt %in% c("Curveball","CurveBall")                              ~ "Curveball",
-    pt == "Slider"                                                   ~ "Slider",
-    pt == "Sweeper"                                                  ~ "Sweeper",
-    pt %in% c("ChangeUp","Changeup")                                ~ "Changeup",
-    pt == "Splitter"                                                 ~ "Splitter",
-    TRUE                                                             ~ NA_character_
-  )
-}
-
-# ── Stuff+ ─────────────────────────────────────────────────────────────────────
-compute_stuff_plus <- function(data) {
-  fb_avgs <- data %>%
-    filter(TaggedPitchType %in% c("Fastball","FourSeamFastBall","Four-Seam","FourSeam",
-                                  "Sinker","TwoSeamFastBall","OneSeamFastball","Cutter")) %>%
-    mutate(
-      PitcherId = as.character(PitcherId),
-      priority  = case_when(
-        TaggedPitchType %in% c("Fastball","FourSeamFastBall","Four-Seam","FourSeam") ~ 1,
-        TaggedPitchType %in% c("Sinker","TwoSeamFastBall","OneSeamFastball")         ~ 2,
-        TaggedPitchType == "Cutter"                                                  ~ 3
-      )
-    ) %>%
-    group_by(PitcherId, priority) %>%
-    summarize(count=n(), fb_velo=mean(RelSpeed,na.rm=TRUE),
-              fb_ivb=mean(InducedVertBreak,na.rm=TRUE),
-              fb_hb=mean(HorzBreak,na.rm=TRUE), .groups="drop") %>%
-    arrange(PitcherId, priority, desc(count)) %>%
-    slice_head(by=PitcherId, n=1) %>%
-    select(PitcherId, PrimaryFB=priority, fb_velo, fb_ivb, fb_hb) %>%
-    mutate(PrimaryFB=case_when(
-      PrimaryFB==1 ~ "Fastball", PrimaryFB==2 ~ "Sinker", PrimaryFB==3 ~ "Cutter"
-    ))
-
-  if (nrow(fb_avgs)==0) return(tibble(Pitcher=character(), `Stuff+`=numeric()))
-
-  scored <- data %>%
-    mutate(PitcherId=as.character(PitcherId),
-           PitchType=map_pitch_type(TaggedPitchType)) %>%
-    filter(!is.na(PitchType)) %>%
-    left_join(fb_avgs, by="PitcherId") %>%
-    mutate(
-      FF_diff  = ifelse(PitchType==PrimaryFB, 0, RelSpeed-fb_velo),
-      IVB_diff = ifelse(PitchType==PrimaryFB, 0, InducedVertBreak-fb_ivb),
-      HB_diff  = ifelse(PitchType==PrimaryFB, 0, HorzBreak-fb_hb),
-      SameSide = as.factor(ifelse(BatterSide==PitcherThrows, 1, 0)),
-      PitchType= as.factor(PitchType)
-    ) %>%
-    filter(!is.na(InducedVertBreak), !is.na(RelSide), !is.na(HorzBreak),
-           !is.na(SpinRate), !is.na(Extension), !is.na(RelHeight),
-           !is.na(RelSpeed), !is.na(SameSide))
-
-  if (nrow(scored)==0) return(tibble(Pitcher=character(), `Stuff+`=numeric()))
-
-  scored %>%
-    bind_cols(suppressWarnings(predict(model, scored, type="prob"))) %>%
-    rename(expected_whiff=.pred_1) %>%
-    group_by(Pitcher, PitchType) %>%
-    summarise(expected_whiff_rate=mean(expected_whiff), .groups="drop") %>%
-    left_join(league_stats, by="PitchType") %>%
-    group_by(Pitcher) %>%
-    summarise(
-      `Stuff+` = round(weighted.mean(
-        ((expected_whiff_rate - league_avg) / league_sd) * 10 + 100,
-        w=rep(1,n()), na.rm=TRUE), 0),
-      .groups="drop"
-    )
-}
-
-# ── Location+ ──────────────────────────────────────────────────────────────────
-compute_location_plus <- function(data) {
-  loc_data <- data %>%
-    mutate(
-      Balls       = as.integer(Balls),
-      Strikes     = as.integer(Strikes),
-      count_state = paste0(Balls,"-",Strikes),
-      count_state = ifelse(!count_state %in% c("0-0","0-1","0-2","1-0","1-1","1-2",
-                                               "2-0","2-1","2-2","3-0","3-1","3-2"),
-                           NA, count_state),
-      count_state = as.factor(count_state),
-      PitchType   = map_pitch_type(TaggedPitchType)
-    ) %>%
-    filter(!is.na(PitchType), !is.na(count_state))
-
-  if (nrow(loc_data)==0) return(tibble(Pitcher=character(), `Location+`=numeric()))
-
-  predictions <- predict(xgb_fit, new_data=loc_data) %>% rename(xRV288=.pred)
-
-  bind_cols(predictions, loc_data) %>%
-    group_by(Pitcher, PitchType) %>%
-    summarise(avg_xRV288=mean(xRV288,na.rm=TRUE), .groups="drop") %>%
-    left_join(league_stats_pitcher, by="PitchType") %>%
-    group_by(Pitcher) %>%
-    summarise(
-      `Location+` = round(weighted.mean(
-        (-(avg_xRV288 - league_avg) / league_sd) * 10 + 100,
-        w=rep(1,n()), na.rm=TRUE), 0),
-      .groups="drop"
-    )
-}
-
 # ── Pitcher stats ──────────────────────────────────────────────────────────────
-build_pitcher_stats <- function(df, min_pitches, stuff_scores, loc_scores) {
+build_pitcher_stats <- function(df, min_pitches, scores) {
 
   team_lookup <- df %>%
     select(Pitcher, PitcherTeam) %>%
@@ -207,9 +99,8 @@ build_pitcher_stats <- function(df, min_pitches, stuff_scores, loc_scores) {
                         na.rm=TRUE),
       .groups     = "drop"
     ) %>%
-    left_join(team_lookup,  by="Pitcher") %>%
-    left_join(stuff_scores, by="Pitcher") %>%
-    left_join(loc_scores,   by="Pitcher") %>%
+    left_join(team_lookup, by="Pitcher") %>%
+    left_join(scores %>% select(Pitcher, `Stuff+`, `Location+`), by="Pitcher") %>%
     filter(Pitches >= min_pitches) %>%
     mutate(
       Name        = sapply(Pitcher, format_name),
@@ -307,7 +198,8 @@ ui <- fluidPage(
       .modal-subtitle { font-size:12px; color:#888; margin:0 0 16px 0; }
       .section-label  { font-size:12px; font-weight:700; color:#002147;
                         text-transform:uppercase; letter-spacing:0.4px;
-                        margin:16px 0 8px 0; border-top:1px solid #eee; padding-top:14px; }
+                        margin:16px 0 8px 0; border-top:1px solid #eee;
+                        padding-top:14px; }
 
       #splash { position:fixed; top:0; left:0; width:100%; height:100%;
                 background:#002147; display:flex; flex-direction:column;
@@ -428,21 +320,12 @@ ui <- fluidPage(
 # ── Server ─────────────────────────────────────────────────────────────────────
 server <- function(input, output, session) {
 
-  # load data + compute models once per level change
   raw_data <- reactive({
     req(input$level)
     tryCatch({
-      df <- load_data(input$level)
       list(
-        df           = df,
-        stuff_scores = tryCatch(
-          compute_stuff_plus(df),
-          error=function(e) tibble(Pitcher=character(), `Stuff+`=numeric())
-        ),
-        loc_scores   = tryCatch(
-          compute_location_plus(df),
-          error=function(e) tibble(Pitcher=character(), `Location+`=numeric())
-        )
+        df     = load_data(input$level),
+        scores = load_scores(input$level)
       )
     }, error = function(e) {
       showNotification(paste("Failed to load:", e$message), type="error")
@@ -456,8 +339,7 @@ server <- function(input, output, session) {
       build_pitcher_stats(
         raw_data()$df,
         input$min_pitches,
-        raw_data()$stuff_scores,
-        raw_data()$loc_scores
+        raw_data()$scores
       )
   })
 
@@ -513,7 +395,6 @@ server <- function(input, output, session) {
     )
   })
 
-  # modal
   current_pitcher <- reactiveVal(NULL)
 
   observeEvent(input$selected_pitcher, {
@@ -548,7 +429,7 @@ server <- function(input, output, session) {
 
   output$movement_plot <- renderPlot({
     req(pitcher_plot_data(), nrow(pitcher_plot_data()) > 0)
-    pdata <- pitcher_plot_data()
+    pdata        <- pitcher_plot_data()
     parts        <- strsplit(current_pitcher(), ", ")[[1]]
     display_name <- if (length(parts)==2) paste(trimws(parts[2]),trimws(parts[1])) else current_pitcher()