Update dataset loaders

nx-datasets/example/california.ml

@@ -15,8 +15,14 @@ let () =
   let labels_1d = Nx.reshape [| n_samples |] labels in
   let labels_f32 = astype_f32 labels_1d in
 
-  let longitude = Nx.slice [ Nx.R (0, n_samples); Nx.R (0, 1) ] features in
-  let latitude = Nx.slice [ Nx.R (0, n_samples); Nx.R (1, 2) ] features in
+  (* slice produces shape [n;1] — reshape to 1-D [n] so Hugin's scatter receives
+     a vector not a 2-D column *)
+  let longitude_col = Nx.slice [ Nx.R (0, n_samples); Nx.R (0, 1) ] features in
+  let latitude_col = Nx.slice [ Nx.R (0, n_samples); Nx.R (1, 2) ] features in
+
+  let longitude = Nx.reshape [| n_samples |] longitude_col in
+  let latitude = Nx.reshape [| n_samples |] latitude_col in
+
   let longitude_f32 = astype_f32 longitude in
   let latitude_f32 = astype_f32 latitude in
 

nx-datasets/lib/datasets/airline_passengers.ml

@@ -10,11 +10,18 @@ let data_path = dataset_dir ^ data_filename
 let url =
   "https://raw.githubusercontent.com/jbrownlee/Datasets/master/airline-passengers.csv"
 
+(* Logging source for this loader *)
+let src =
+  Logs.Src.create "nx.datasets.airline_passengers"
+    ~doc:"Airline passengers loader"
+
+module Log = (val Logs.src_log src : Logs.LOG)
+
 let ensure_dataset () = ensure_file url data_path
 
 let load () =
   ensure_dataset ();
-  Printf.printf "Loading Airline Passengers dataset...\n%!";
+  Log.info (fun m -> m "Loading Airline Passengers dataset...");
 
   let header, data_rows_iter =
     try
@@ -58,9 +65,11 @@ let load () =
           let row_list = Csv.Row.to_list row in
           (* Convert Row.t to string list *)
           if List.length row_list <> List.length header then
-            Printf.eprintf "Warning: Row %d has %d columns, expected %d\n%!"
-              (List.length acc + 1)
-              (List.length row_list) (List.length header);
+            Log.warn (fun m ->
+                m "Row %d has %d columns, expected %d (header: %s)"
+                  (List.length acc + 1)
+                  (List.length row_list) (List.length header)
+                  (String.concat ", " header));
 
           (* Check length before accessing *)
           if List.length row_list > passenger_col_index then
@@ -73,12 +82,13 @@ let load () =
             let passenger_int = parse_int_cell ~context passenger_str in
             passenger_int :: acc
           else (
-            Printf.eprintf
-              "Warning: Row %d is shorter than expected (%d < %d), skipping \
-               passenger value.\n\
-               %!"
-              (List.length acc + 1)
-              (List.length row_list) (passenger_col_index + 1);
+            Log.warn (fun m ->
+                m
+                  "Row %d is shorter than expected (%d < %d), skipping \
+                   passenger value. Missing column: %s"
+                  (List.length acc + 1)
+                  (List.length row_list) (passenger_col_index + 1)
+                  passenger_col_name);
             -1 :: acc (* Placeholder for missing data *)))
         ~init:[] data_rows_iter
     with
@@ -101,7 +111,7 @@ let load () =
   let num_samples = List.length data_rows_rev in
   if num_samples = 0 then
     failwith "No data rows loaded from airline-passengers.csv";
-  Printf.printf "Found %d samples.\n%!" num_samples;
+  Log.info (fun m -> m "Found %d samples." num_samples);
 
   (* Create Bigarray and populate (data is reversed from fold_left) *)
   let passengers = Array1.create int32 c_layout num_samples in
@@ -110,5 +120,5 @@ let load () =
       passengers.{num_samples - 1 - i} <- Int32.of_int passenger_val)
     data_rows_rev;
 
-  Printf.printf "Airline Passengers loading complete.\n%!";
+  Log.info (fun m -> m "Airline Passengers loading complete.");
   passengers

nx-datasets/lib/datasets/breast_cancer.ml

@@ -2,6 +2,13 @@
 open Bigarray
 open Dataset_utils
 
+(* Logging source for this loader *)
+let src =
+  Logs.Src.create "nx.datasets.breast_cancer"
+    ~doc:"Breast Cancer dataset loader"
+
+module Log = (val Logs.src_log src : Logs.LOG)
+
 let dataset_name = "breast-cancer-wisconsin"
 let dataset_dir = get_cache_dir dataset_name
 let data_filename = "wdbc.data"
@@ -24,7 +31,7 @@ let encode_label label row =
 
 let load () =
   ensure_dataset ();
-  Printf.printf "Loading Breast Cancer Wisconsin dataset...\n%!";
+  Log.info (fun m -> m "Loading Breast Cancer Wisconsin dataset...");
 
   let data_rows =
     try
@@ -51,7 +58,7 @@ let load () =
   let expected_cols = 32 in
   let num_features = 30 in
 
-  Printf.printf "Found %d samples.\n%!" num_samples;
+  Log.info (fun m -> m "Found %d samples." num_samples);
 
   let features = Array2.create float64 c_layout num_samples num_features in
   let labels = Array1.create int c_layout num_samples in
@@ -75,5 +82,5 @@ let load () =
       done)
     data_rows;
 
-  Printf.printf "Breast Cancer Wisconsin loading complete.\n%!";
+  Log.info (fun m -> m "Breast Cancer Wisconsin loading complete.");
   (features, labels)

nx-datasets/lib/datasets/california_housing.ml

@@ -2,6 +2,13 @@
 open Bigarray
 open Dataset_utils
 
+(* Logging source for this loader *)
+let src =
+  Logs.Src.create "nx.datasets.california_housing"
+    ~doc:"California Housing loader"
+
+module Log = (val Logs.src_log src : Logs.LOG)
+
 let dataset_name = "california-housing"
 let dataset_dir = get_cache_dir dataset_name
 let data_filename = "housing.csv"
@@ -28,7 +35,7 @@ let calculate_mean_non_nan column_data =
 
 let load () =
   ensure_dataset ();
-  Printf.printf "Loading California Housing dataset...\n%!";
+  Log.info (fun m -> m "Loading California Housing dataset...");
 
   let header, all_data_rows =
     try
@@ -85,8 +92,9 @@ let load () =
     List.find_index (( = ) "total_bedrooms") header
   in
 
-  Printf.printf "Found %d samples. Loading %d features + target '%s'.\n%!"
-    num_samples num_features target_name;
+  Log.info (fun m ->
+      m "Found %d samples. Loading %d features + target '%s'." num_samples
+        num_features target_name);
 
   let parsed_features_temp = Array.make_matrix num_samples num_features nan in
   let parsed_labels_temp = Array.make num_samples nan in
@@ -95,8 +103,10 @@ let load () =
   List.iteri
     (fun i row ->
       if List.length row <> List.length header then
-        Printf.eprintf "Warning: Row %d has %d columns, expected %d\n%!" (i + 1)
-          (List.length row) (List.length header);
+        Log.warn (fun m ->
+            m "Row %d has %d columns, expected %d (header: %s)" (i + 1)
+              (List.length row) (List.length header)
+              (String.concat ", " header));
 
       List.iteri
         (fun j feature_idx ->
@@ -106,29 +116,30 @@ let load () =
             parsed_features_temp.(i).(j) <- v_float;
             if Some feature_idx = total_bedrooms_index_opt then
               total_bedrooms_col_temp := v_float :: !total_bedrooms_col_temp)
-          else (
-            Printf.eprintf
-              "Warning: Row %d missing feature column %d ('%s'). Setting NaN.\n\
-               %!"
-              (i + 1) feature_idx (List.nth feature_names j);
+          else
+            let feature_name = List.nth feature_names j in
+            Log.warn (fun m ->
+                m "Row %d missing feature column %d ('%s'). Setting NaN."
+                  (i + 1) feature_idx feature_name);
             parsed_features_temp.(i).(j) <- nan;
             if Some feature_idx = total_bedrooms_index_opt then
-              total_bedrooms_col_temp := nan :: !total_bedrooms_col_temp))
+              total_bedrooms_col_temp := nan :: !total_bedrooms_col_temp)
         feature_indices;
 
       if List.length row > target_index then
         let label_str = List.nth row target_index in
         parsed_labels_temp.(i) <- parse_float_or_nan label_str
       else (
-        Printf.eprintf
-          "Warning: Row %d missing target column %d ('%s'). Setting NaN.\n%!"
-          (i + 1) target_index target_name;
+        Log.warn (fun m ->
+            m "Row %d missing target column %d ('%s'). Setting NaN." (i + 1)
+              target_index target_name);
         parsed_labels_temp.(i) <- nan))
     data_rows_str;
 
   let total_bedrooms_mean = calculate_mean_non_nan !total_bedrooms_col_temp in
-  Printf.printf "Calculated mean for 'total_bedrooms' (for imputation): %f\n%!"
-    total_bedrooms_mean;
+  Log.info (fun m ->
+      m "Calculated mean for 'total_bedrooms' (for imputation): %f"
+        total_bedrooms_mean);
   let total_bedrooms_feature_index =
     match List.find_index (( = ) "total_bedrooms") feature_names with
     | Some idx -> idx
@@ -163,5 +174,5 @@ let load () =
     else labels.{i} <- label_v
   done;
 
-  Printf.printf "California Housing loading complete.\n%!";
+  Log.info (fun m -> m "California Housing loading complete.");
   (features, labels)

nx-datasets/lib/datasets/cifar10.ml

@@ -10,12 +10,17 @@ let url = "https://www.cs.toronto.edu/~kriz/cifar-10-python.tar.gz"
 let tar_path = base_dir ^ Filename.basename url
 let test_batch_rel_path = archive_dir_name ^ "/test_batch"
 
+(* Logging source for this loader *)
+let src = Logs.Src.create "nx.datasets.cifar10" ~doc:"CIFAR10 dataset loader"
+
+module Log = (val Logs.src_log src : Logs.LOG)
+
 let ensure_dataset () =
   ensure_extracted_archive ~url ~archive_path:tar_path ~extract_dir:base_dir
     ~check_file:test_batch_rel_path
 
 let read_cifar_batch filename =
-  Printf.printf "Reading batch file: %s\n%!" filename;
+  Log.debug (fun m -> m "Reading batch file: %s" filename);
   let ic = open_in_bin filename in
   let s =
     try really_input_string ic (in_channel_length ic)
@@ -32,7 +37,7 @@ let read_cifar_batch filename =
       (Printf.sprintf "File %s has unexpected size %d" filename num_bytes);
 
   let num_images = num_bytes / bytes_per_image in
-  Printf.printf "Found %d images in %s.\n%!" num_images filename;
+  Log.debug (fun m -> m "Found %d images in %s." num_images filename);
 
   let images =
     Genarray.create int8_unsigned c_layout [| num_images; 32; 32; 3 |]
@@ -64,7 +69,7 @@ let read_cifar_batch filename =
 
 let load () =
   ensure_dataset ();
-  Printf.printf "Loading CIFAR-10 dataset...\n%!";
+  Log.info (fun m -> m "Loading CIFAR-10 dataset...");
 
   let train_batches_files =
     List.init 5 (fun i -> dataset_dir ^ Printf.sprintf "data_batch_%d" (i + 1))
@@ -107,5 +112,5 @@ let load () =
   let test_batch_file = dataset_dir ^ "test_batch" in
   let test_images, test_labels = read_cifar_batch test_batch_file in
 
-  Printf.printf "CIFAR-10 loading complete.\n%!";
+  Log.info (fun m -> m "CIFAR-10 loading complete.");
   ((train_images, train_labels), (test_images, test_labels))

nx-datasets/lib/datasets/diabetes.ml

@@ -2,6 +2,11 @@
 open Bigarray
 open Dataset_utils
 
+(* Logging source for this loader *)
+let src = Logs.Src.create "nx.datasets.diabetes" ~doc:"Diabetes dataset loader"
+
+module Log = (val Logs.src_log src : Logs.LOG)
+
 let dataset_name = "diabetes-sklearn"
 let dataset_dir = get_cache_dir dataset_name
 let data_filename = "diabetes.tab.txt"
@@ -11,7 +16,7 @@ let ensure_dataset () = ensure_file url data_path
 
 let load () =
   ensure_dataset ();
-  Printf.printf "Loading Diabetes (sklearn version) dataset...\n%!";
+  Log.info (fun m -> m "Loading Diabetes (sklearn version) dataset...");
 
   let header, data_rows_iter =
     try
@@ -106,8 +111,9 @@ let load () =
   let num_samples = List.length labels_rev in
 
   if num_samples = 0 then failwith "No data rows loaded from diabetes.tab.txt";
-  Printf.printf "Found %d samples with %d features and target '%s'.\n%!"
-    num_samples num_features target_col_name;
+  Log.info (fun m ->
+      m "Found %d samples with %d features and target '%s'." num_samples
+        num_features target_col_name);
 
   let features_ba = Array2.create float64 c_layout num_samples num_features in
   let labels_ba = Array1.create float64 c_layout num_samples in
@@ -124,5 +130,5 @@ let load () =
     (fun i label_val -> labels_ba.{num_samples - 1 - i} <- label_val)
     labels_rev;
 
-  Printf.printf "Diabetes loading complete.\n%!";
+  Log.info (fun m -> m "Diabetes loading complete.");
   (features_ba, labels_ba)

nx-datasets/lib/datasets/iris.ml

@@ -2,6 +2,10 @@
 open Bigarray
 open Dataset_utils
 
+let src = Logs.Src.create "nx.datasets.iris" ~doc:"Iris dataset loader"
+
+module Log = (val Logs.src_log src : Logs.LOG)
+
 let dataset_name = "iris"
 let dataset_dir = get_cache_dir dataset_name
 let data_filename = "iris.data"
@@ -24,7 +28,7 @@ let encode_label s =
 
 let load () =
   ensure_dataset ();
-  Printf.printf "Loading Iris dataset...\n%!";
+  Log.info (fun m -> m "Loading Iris dataset...");
 
   let data_rows =
     try
@@ -50,7 +54,7 @@ let load () =
   let num_features = 4 in
 
   if num_samples = 0 then failwith "No data loaded from iris.data";
-  Printf.printf "Found %d samples.\n%!" num_samples;
+  Log.info (fun m -> m "Found %d samples" num_samples);
 
   let features = Array2.create float64 c_layout num_samples num_features in
   let labels = Array1.create int32 c_layout num_samples in
@@ -73,5 +77,5 @@ let load () =
       labels.{i} <- encode_label label_str)
     data_rows;
 
-  Printf.printf "Iris loading complete.\n%!";
+  Log.info (fun m -> m "Iris loading complete");
   (features, labels)