Fix dataset structure for proper HF viewer display

- Added dataset loading script (abscorrespondance.py)
- Created data/train.parquet with 95,674 master correspondence records
- Removed branch_catchments files that confused the viewer
- Added dataset_info.json for metadata

Now the full 95,674-record master correspondence table should display properly.

abscorrespondance.py

@@ -0,0 +1,70 @@
+"""Australian Geographic Correspondence Table dataset."""
+
+import datasets
+import pandas as pd
+
+_DESCRIPTION = """
+A comprehensive correspondence table linking Australian geographic hierarchies (POA ↔ LGA ↔ SA2 ↔ Branch Catchments) 
+built using real Australian Bureau of Statistics (ABS) data. Contains 95,674 records with spatial intersection weights.
+"""
+
+_HOMEPAGE = "https://github.com/Mrassimo/HCFcorrtable"
+
+_LICENSE = "MIT"
+
+_URLS = {
+    "train": "master_correspondence_table.parquet",
+}
+
+
+class ABSCorrespondance(datasets.GeneratorBasedBuilder):
+    """Australian Geographic Correspondence Table dataset."""
+
+    VERSION = datasets.Version("1.0.0")
+
+    def _info(self):
+        features = datasets.Features({
+            "poa_code": datasets.Value("string"),
+            "poa_name": datasets.Value("int64"),  
+            "lga_code": datasets.Value("string"),
+            "lga_name": datasets.Value("string"),
+            "sa2_code": datasets.Value("string"),
+            "sa2_name": datasets.Value("string"),
+            "branch_id": datasets.Value("string"),
+            "branch_name": datasets.Value("string"),
+            "branch_type": datasets.Value("string"),
+            "branch_lat": datasets.Value("float64"),
+            "branch_long": datasets.Value("float64"),
+            "poa_to_lga_weight": datasets.Value("float64"),
+            "lga_to_sa2_weight": datasets.Value("float64"),
+            "catchment_overlap_pct": datasets.Value("float64"),
+            "distance_to_branch_km": datasets.Value("float64"),
+            "record_type": datasets.Value("string"),
+        })
+        
+        return datasets.DatasetInfo(
+            description=_DESCRIPTION,
+            features=features,
+            homepage=_HOMEPAGE,
+            license=_LICENSE,
+        )
+
+    def _split_generators(self, dl_manager):
+        urls = _URLS
+        data_path = dl_manager.download(urls["train"])
+        
+        return [
+            datasets.SplitGenerator(
+                name=datasets.Split.TRAIN,
+                gen_kwargs={
+                    "filepath": data_path,
+                },
+            ),
+        ]
+
+    def _generate_examples(self, filepath):
+        """Yields examples from the dataset."""
+        df = pd.read_parquet(filepath)
+        
+        for idx, row in df.iterrows():
+            yield idx, row.to_dict()

branch_catchments.parquet

@@ -1,3 +0,0 @@
-version https://git-lfs.github.com/spec/v1
-oid sha256:e671cd29e6ab78e6c8ddaf7e7bc1dfdc5be8847397838f870bfd335f9d71e3b2
-size 39183

branch_catchments.csv → data/train.parquet

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:f4b29705cf5fe9451b5cf4a818f02b3eaa3a9fa5fcc685d0fa88fe226cd79900
-size 38512
+oid sha256:f6f2eb58ceba119bc399dbb758728f59bc17d9ccde1b636a83156923aa48e0db
+size 277194

dataset_info.json

@@ -0,0 +1,64 @@
+{
+  "dataset_name": "abscorrespondance",
+  "dataset_size": 95674,
+  "features": {
+    "poa_code": {
+      "dtype": "int64"
+    },
+    "poa_name": {
+      "dtype": "int64"
+    },
+    "lga_code": {
+      "dtype": "int64"
+    },
+    "lga_name": {
+      "dtype": "object"
+    },
+    "sa2_code": {
+      "dtype": "int64"
+    },
+    "sa2_name": {
+      "dtype": "object"
+    },
+    "branch_id": {
+      "dtype": "object"
+    },
+    "branch_name": {
+      "dtype": "object"
+    },
+    "branch_type": {
+      "dtype": "object"
+    },
+    "branch_lat": {
+      "dtype": "float64"
+    },
+    "branch_long": {
+      "dtype": "float64"
+    },
+    "poa_to_lga_weight": {
+      "dtype": "float64"
+    },
+    "lga_to_sa2_weight": {
+      "dtype": "float64"
+    },
+    "catchment_overlap_pct": {
+      "dtype": "float64"
+    },
+    "distance_to_branch_km": {
+      "dtype": "float64"
+    },
+    "record_type": {
+      "dtype": "object"
+    }
+  },
+  "splits": {
+    "train": {
+      "name": "train",
+      "num_examples": 95674,
+      "dataset_name": "abscorrespondance",
+      "data_files": [
+        "data/train.parquet"
+      ]
+    }
+  }
+}