Update app.R

app.R

@@ -2241,7 +2241,7 @@ has_col_index <- function(idx) {
 
             
 stuffplus_model <- tryCatch({
-  readr::read_rds("college_stuffplusCC.rds")
+  readr::read_rds("new_stuffplus_model.rds")
 }, error = function(e) {
   message("Stuff+ model not found. Predictions will be unavailable.")
   NULL
@@ -2299,68 +2299,40 @@ preflight_predictor_check <- function(model, newdata) {
 }
 
 ensure_stuff_inputs <- function(df) {
-  core <- c("ZoneTime","ax0","az0","RelHeight","RelSide","SpinRate","SpinAxis","Extension")
-  missing_core <- setdiff(core, names(df))
-  if (length(missing_core)) {
-    message("Stuff+ required raw columns missing: ", paste(missing_core, collapse = ", "))
-  }
-  
-  if ("TaggedPitchType" %in% names(df) && nrow(df) > 0) {
-    pitch_counts <- df %>%
-      filter(!is.na(TaggedPitchType), TaggedPitchType != "Other") %>%
-      count(TaggedPitchType, sort = TRUE)
-    
-    if (nrow(pitch_counts) > 0) {
-      primary_pitch <- pitch_counts$TaggedPitchType[1]
-      message("Primary pitch detected: ", primary_pitch)
-      
-      primary_velo <- mean(df$RelSpeed[df$TaggedPitchType == primary_pitch], na.rm = TRUE)
-      primary_az0 <- mean(df$az0[df$TaggedPitchType == primary_pitch], na.rm = TRUE)
-      
-      if (!"velo_diff" %in% names(df)) {
-        if ("RelSpeed" %in% names(df)) {
-          if (is.finite(primary_velo)) {
-            df$velo_diff <- df$RelSpeed - primary_velo
-            message("velo_diff calculated relative to primary pitch (mean=", round(primary_velo, 2), ")")
-          } else {
-            message("Primary velo not finite; setting velo_diff = 0")
-            df$velo_diff <- 0
-          }
-        } else {
-          message("RelSpeed not found; setting velo_diff = 0 as fallback")
-          df$velo_diff <- 0
-        }
-      }
-      
-      if (!"az0_diff" %in% names(df)) {
-        if ("az0" %in% names(df)) {
-          if (is.finite(primary_az0)) {
-            df$az0_diff <- df$az0 - primary_az0
-            message("az0_diff calculated relative to primary pitch (mean=", round(primary_az0, 2), ")")
-          } else {
-            message("Primary az0 not finite; setting az0_diff = 0")
-            df$az0_diff <- 0
-          }
-        } else {
-          message("az0 not found; setting az0_diff = 0 as fallback")
-          df$az0_diff <- 0
-        }
-      }
-    } else {
-      message("No valid pitch types found; setting engineered features = 0")
-      if (!"velo_diff" %in% names(df)) df$velo_diff <- 0
-      if (!"az0_diff" %in% names(df)) df$az0_diff <- 0
-    }
-  } else {
-    message("No TaggedPitchType column; setting engineered features = 0")
-    if (!"velo_diff" %in% names(df)) df$velo_diff <- 0
-    if (!"az0_diff" %in% names(df)) df$az0_diff <- 0
-  }
   
   df <- df %>% 
-    mutate(ax0 = -ax0)
+    mutate(RelSide = case_when(
+      PitcherThrows == "Right" ~ RelSide,
+      PitcherThrows == "Left" ~ -RelSide,
+      PitcherThrows %in% c("Both", "Undefined") & RelSide > 0 ~ RelSide,
+      PitcherThrows %in% c("Both", "Undefined") & RelSide < 0 ~ -RelSide),
+          ax0 = case_when(
+            PitcherThrows == "Right" ~ ax0,
+            PitcherThrows == "Left" ~ -ax0,
+            PitcherThrows %in% c("Both", "Undefined") & ax0 > 0 ~ ax0,
+            PitcherThrows %in% c("Both", "Undefined") & ax0 < 0 ~ -ax0),
+                 PlateLocHeight = PlateLocHeight*12,
+                 PlateLocSide = PlateLocSide*12,
+          ax0 = -ax0) %>%
+    group_by(Pitcher, GameID) %>%
+    mutate(
+      primary_pitch = case_when(
+        any(TaggedPitchType == "Fastball") ~ "Fastball",
+        any(TaggedPitchType == "Sinker") ~ "Sinker",
+        TRUE ~ names(sort(table(TaggedPitchType), decreasing = TRUE))[1]
+      )
+    ) %>%
+    group_by(Pitcher, GameID, primary_pitch) %>%
+    mutate(
+      primary_az0 = mean(az0[TaggedPitchType == primary_pitch], na.rm = TRUE),
+      primary_velo = mean(RelSpeed[TaggedPitchType == primary_pitch], na.rm = TRUE)
+    ) %>%
+    ungroup() %>% 
+    mutate(az0_diff = az0 - primary_az0, 
+           velo_diff = RelSpeed - primary_velo)
   
   df
+  
 }
 
 standardize_stuffplus_to_league <- function(data, league_comparison_data) {