finalising

Urban-Analytics-Technology-Platform · Dec 19, 2023 · cc11fc1 · cc11fc1
1 parent 9debb33
commit cc11fc1
Showing 1 changed file with 52 additions and 12 deletions.
diff --git a/code/04_generalisation/prototype.ipynb b/code/04_generalisation/prototype.ipynb
@@ -72,11 +72,12 @@
   },
   {
    "cell_type": "code",
-   "execution_count": 3,
+   "execution_count": 81,
    "metadata": {},
    "outputs": [],
    "source": [
-    "name = \"newcastle_h3\"\n",
+    "name = \"tyne_and_wear_hex\" # used for files\n",
+    "area_name = \"Tyne and Wear\" # used in the app\n",
     "aoi_file_path = \"https://github.com/Urban-Analytics-Technology-Platform/demoland-web/raw/main/web/src/data/geography.json\"\n",
     "gtfs_data_file_path = f\"{data_folder}/raw/accessibility/itm_north_east_gtfs.zip\""
    ]
@@ -99,6 +100,7 @@
     "os.mkdir(\"app\")\n",
     "os.mkdir(\"app/scenarios\")\n",
     "os.mkdir(\"engine\")\n",
+    "os.mkdir(f\"engine/{name}\")\n",
     "os.mkdir(\"temp\")"
    ]
   },
@@ -181,6 +183,15 @@
     "grid_aoi[[\"lat\", \"lon\"]] = pd.DataFrame(grid_aoi.index.to_series().apply(h3.h3_to_geo).tolist(), columns=[\"lat\", \"lon\"], index=grid_aoi.index)"
    ]
   },
+  {
+   "cell_type": "code",
+   "execution_count": 70,
+   "metadata": {},
+   "outputs": [],
+   "source": [
+    "grid_aoi = grid_aoi.dropna(subset=\"signature_type\")"
+   ]
+  },
   {
    "cell_type": "markdown",
    "metadata": {},
@@ -214,7 +225,7 @@
    "metadata": {},
    "outputs": [],
    "source": [
-    "matrix_aoi.to_parquet(\"engine/matrix.parquet\")"
+    "matrix_aoi.to_parquet(f\"engine/{name}/matrix.parquet\")"
    ]
   },
   {
@@ -908,7 +919,7 @@
    "metadata": {},
    "outputs": [],
    "source": [
-    "with open(f\"engine/accessibility.joblib\", \"wb\") as f:\n",
+    "with open(f\"engine/{name}/accessibility.joblib\", \"wb\") as f:\n",
     "    joblib.dump(acc, f, compress=True)"
    ]
   },
@@ -923,11 +934,40 @@
   },
   {
    "cell_type": "code",
-   "execution_count": 49,
+   "execution_count": 65,
+   "metadata": {},
+   "outputs": [],
+   "source": [
+    "grid_aoi.geometry.to_crs(4326).to_frame().assign(id=range(len(grid_aoi))).to_file(\"app/geography.json\")"
+   ]
+  },
+  {
+   "cell_type": "markdown",
+   "metadata": {},
+   "source": [
+    "Geo config"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 82,
    "metadata": {},
    "outputs": [],
    "source": [
-    "grid_aoi.geometry.to_crs(4326).to_file(\"app/geography.json\")"
+    "bds = grid_aoi.to_crs(4326).total_bounds\n",
+    "zoom_lon = np.log2(360 * 2.0 / (bds[2] - bds[0]))\n",
+    "zoom_lat = np.log2(360 * 2.0 / (bds[3] - bds[1]))\n",
+    "zoom = round(np.min([zoom_lon, zoom_lat]), 2)\n",
+    "geo_config = {\n",
+    "    \"featureIdentifier\": \"to_id\",\n",
+    "    \"initialLatitude\": round(np.mean([bds[1], bds[3]]), 2),\n",
+    "    \"initialLongitude\": round(np.mean([bds[0], bds[2]]), 2),\n",
+    "    \"initialZoom\": zoom - .4,\n",
+    "    \"areaName\": area_name,\n",
+    "    \"modelIdentifier\": name,\n",
+    "}\n",
+    "with open(\"app/geo_config.json\", \"w\") as f:\n",
+    "    json.dump(geo_config, f)"
    ]
   },
   {
@@ -978,7 +1018,7 @@
    "metadata": {},
    "outputs": [],
    "source": [
-    "with open(f\"engine/accessibility.joblib\", \"rb\") as f:\n",
+    "with open(f\"engine/{name}/accessibility.joblib\", \"rb\") as f:\n",
     "    acc = joblib.load(f)"
    ]
   },
@@ -1068,7 +1108,7 @@
     "    \"air_quality_index\",\n",
     "    \"house_price_index\",\n",
     "    \"signature_type\"\n",
-    "]).to_parquet(\"engine/default_data.parquet\")"
+    "]).to_parquet(f\"engine/{name}/default_data.parquet\")"
    ]
   },
   {
@@ -1077,7 +1117,7 @@
    "metadata": {},
    "outputs": [],
    "source": [
-    "pd.DataFrame(index=grid_aoi.index, columns=[\"signature_type\",\"use\", \"greenspace\", \"job_types\"]).replace(np.nan, None).to_parquet(\"engine/empty.parquet\")"
+    "pd.DataFrame(index=grid_aoi.index, columns=[\"signature_type\",\"use\", \"greenspace\", \"job_types\"]).replace(np.nan, None).to_parquet(f\"engine/{name}/empty.parquet\")"
    ]
   },
   {
@@ -1086,7 +1126,7 @@
    "metadata": {},
    "outputs": [],
    "source": [
-    "pd.DataFrame({\"area\": grid_aoi.area}).to_parquet(\"engine/oa_area.parquet\")"
+    "pd.DataFrame({\"area\": grid_aoi.area}).to_parquet(f\"engine/{name}/oa_area.parquet\")"
    ]
   },
   {
@@ -1095,7 +1135,7 @@
    "metadata": {},
    "outputs": [],
    "source": [
-    "grid_aoi[[\"signature_type\"]].rename(columns={\"signature_type\": \"primary_type\"}).to_parquet(\"engine/oa_key.parquet\")"
+    "grid_aoi[[\"signature_type\"]].rename(columns={\"signature_type\": \"primary_type\"}).to_parquet(f\"engine/{name}/oa_key.parquet\")"
    ]
   },
   {
@@ -1112,7 +1152,7 @@
    "outputs": [],
    "source": [
     "registry = {}\n",
-    "for fp in glob(\"engine/*\"):\n",
+    "for fp in glob(f\"engine/{name}/\"):\n",
     "    with open(fp, \"rb\") as f:\n",
     "        bytes = f.read()\n",
     "        registry[fp] = hashlib.sha256(bytes).hexdigest()\n",