Update app.R

app.R

@@ -4,6 +4,8 @@ library(arrow)
 library(DT)
 library(ggplot2)
 library(shinyjs)
+library(tidymodels)
+library(xgboost)
 
 # ── Load data ──────────────────────────────────────────────────────────────────
 load_data <- function(level) {
@@ -11,6 +13,12 @@ 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")
+
 # ── Pitch colors ───────────────────────────────────────────────────────────────
 pitch_colors <- c(
   "Fastball"  = "red",       "Sinker"   = "orange",
@@ -33,6 +41,112 @@ clean_pitch_type <- function(pt) {
   )
 }
 
+# ── Pitch type mapper ──────────────────────────────────────────────────────────
+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) {
 
@@ -94,6 +208,15 @@ build_pitcher_stats <- function(df, min_pitches) {
       .groups     = "drop"
     ) %>%
     left_join(team_lookup, by="Pitcher") %>%
+  %>%
+    left_join(
+      tryCatch(compute_stuff_plus(df),    error=function(e) tibble(Pitcher=character(), `Stuff+`=numeric())),
+      by="Pitcher"
+    ) %>%
+    left_join(
+      tryCatch(compute_location_plus(df), error=function(e) tibble(Pitcher=character(), `Location+`=numeric())),
+      by="Pitcher"
+    ) %>%
     filter(Pitches >= min_pitches) %>%
     mutate(
       Name        = sapply(Pitcher, format_name),
@@ -131,7 +254,7 @@ build_pitcher_stats <- function(df, min_pitches) {
     ) %>%
     select(Name, Level, Team, Pitches, `FB Velo`, `Zone%`, `Strike%`,
            `K's`, `BB's`, `K/BB`, `Whiff%`, `Chase%`, `Hard Hit%`,
-           `Plot`, `Twitter`) %>%
+           `Stuff+`, `Location+`, `Plot`, `Twitter`) %>%
     arrange(desc(Pitches))
 }
 
@@ -390,9 +513,9 @@ server <- function(input, output, session) {
         dom          = "lfrtip",
         columnDefs = list(
           list(className = "dt-left",   targets = 0:1),
-          list(className = "dt-center", targets = 2:14),
-          list(orderable = FALSE,       targets = c(13, 14))
-        ),
+          list(className = "dt-center", targets = 2:16),
+          list(orderable = FALSE,       targets = c(15, 16))
+        )
         initComplete = JS("function(settings, json) {
           $('#loading_msg').hide();
         }")