Update code to fix negative att

haihabi · haihabi · commit 30beca2bb69f · 2025-09-24T15:15:33.000+02:00
diff --git a/examples/example_using_different_dataset.ipynb b/examples/example_using_different_dataset.ipynb
diff --git a/examples/rain_estimation_constant.ipynb b/examples/rain_estimation_constant.ipynb
@@ -11,8 +11,6 @@
   },
   {
    "cell_type": "code",
-   "execution_count": 23,
-   "outputs": [],
    "source": [
     "import pynncml as pnc \n",
     "from matplotlib import pyplot as plt"
@@ -22,8 +20,23 @@
     "pycharm": {
      "name": "#%%\n",
      "is_executing": false
+    },
+    "ExecuteTime": {
+     "end_time": "2025-09-24T13:09:00.957505Z",
+     "start_time": "2025-09-24T13:08:59.407782Z"
     }
-   }
+   },
+   "outputs": [
+    {
+     "name": "stderr",
+     "output_type": "stream",
+     "text": [
+      "/Users/haihab01/envs/research_base/lib/python3.11/site-packages/tqdm/auto.py:21: TqdmWarning: IProgress not found. Please update jupyter and ipywidgets. See https://ipywidgets.readthedocs.io/en/stable/user_install.html\n",
+      "  from .autonotebook import tqdm as notebook_tqdm\n"
+     ]
+    }
+   ],
+   "execution_count": 1
   },
   {
    "cell_type": "markdown",
@@ -36,8 +49,6 @@
   },
   {
    "cell_type": "code",
-   "execution_count": 24,
-   "outputs": [],
    "source": [
     "open_cml_dataset = pnc.read_open_cml_dataset('../dataset/open_cml.p') #  read OpenCML dataset"
    ],
@@ -46,8 +57,26 @@
     "pycharm": {
      "name": "#%% \n",
      "is_executing": false
+    },
+    "ExecuteTime": {
+     "end_time": "2025-09-24T13:09:01.043316Z",
+     "start_time": "2025-09-24T13:09:00.971401Z"
     }
-   }
+   },
+   "outputs": [
+    {
+     "ename": "AttributeError",
+     "evalue": "module 'pynncml' has no attribute 'read_open_cml_dataset'",
+     "output_type": "error",
+     "traceback": [
+      "\u001B[31m---------------------------------------------------------------------------\u001B[39m",
+      "\u001B[31mAttributeError\u001B[39m                            Traceback (most recent call last)",
+      "\u001B[36mCell\u001B[39m\u001B[36m \u001B[39m\u001B[32mIn[2]\u001B[39m\u001B[32m, line 1\u001B[39m\n\u001B[32m----> \u001B[39m\u001B[32m1\u001B[39m open_cml_dataset = \u001B[43mpnc\u001B[49m\u001B[43m.\u001B[49m\u001B[43mread_open_cml_dataset\u001B[49m(\u001B[33m'\u001B[39m\u001B[33m../dataset/open_cml.p\u001B[39m\u001B[33m'\u001B[39m) \u001B[38;5;66;03m#  read OpenCML dataset\u001B[39;00m\n",
+      "\u001B[31mAttributeError\u001B[39m: module 'pynncml' has no attribute 'read_open_cml_dataset'"
+     ]
+    }
+   ],
+   "execution_count": 2
   },
   {
    "cell_type": "markdown",
@@ -205,4 +234,4 @@
  },
  "nbformat": 4,
  "nbformat_minor": 0
-}
+}
diff --git a/examples/rain_estimation_dynamic.ipynb b/examples/rain_estimation_dynamic.ipynb
@@ -9,7 +9,6 @@
   },
   {
    "cell_type": "code",
-   "execution_count": 15,
    "metadata": {
     "collapsed": false,
     "jupyter": {
@@ -18,13 +17,27 @@
     "pycharm": {
      "is_executing": false,
      "name": "#%%\n"
+    },
+    "ExecuteTime": {
+     "end_time": "2025-09-24T13:09:15.625122Z",
+     "start_time": "2025-09-24T13:09:14.128129Z"
     }
    },
-   "outputs": [],
    "source": [
     "import pynncml as pnc \n",
     "from matplotlib import pyplot as plt"
-   ]
+   ],
+   "outputs": [
+    {
+     "name": "stderr",
+     "output_type": "stream",
+     "text": [
+      "/Users/haihab01/envs/research_base/lib/python3.11/site-packages/tqdm/auto.py:21: TqdmWarning: IProgress not found. Please update jupyter and ipywidgets. See https://ipywidgets.readthedocs.io/en/stable/user_install.html\n",
+      "  from .autonotebook import tqdm as notebook_tqdm\n"
+     ]
+    }
+   ],
+   "execution_count": 1
   },
   {
    "cell_type": "markdown",
@@ -35,7 +48,6 @@
   },
   {
    "cell_type": "code",
-   "execution_count": 16,
    "metadata": {
     "collapsed": false,
     "jupyter": {
@@ -44,12 +56,29 @@
     "pycharm": {
      "is_executing": false,
      "name": "#%% \n"
+    },
+    "ExecuteTime": {
+     "end_time": "2025-09-24T13:09:15.707062Z",
+     "start_time": "2025-09-24T13:09:15.635194Z"
     }
    },
-   "outputs": [],
    "source": [
     "open_cml_dataset = pnc.read_open_cml_dataset('../dataset/open_cml.p') #  read OpenCML dataset"
-   ]
+   ],
+   "outputs": [
+    {
+     "ename": "AttributeError",
+     "evalue": "module 'pynncml' has no attribute 'read_open_cml_dataset'",
+     "output_type": "error",
+     "traceback": [
+      "\u001B[31m---------------------------------------------------------------------------\u001B[39m",
+      "\u001B[31mAttributeError\u001B[39m                            Traceback (most recent call last)",
+      "\u001B[36mCell\u001B[39m\u001B[36m \u001B[39m\u001B[32mIn[2]\u001B[39m\u001B[32m, line 1\u001B[39m\n\u001B[32m----> \u001B[39m\u001B[32m1\u001B[39m open_cml_dataset = \u001B[43mpnc\u001B[49m\u001B[43m.\u001B[49m\u001B[43mread_open_cml_dataset\u001B[49m(\u001B[33m'\u001B[39m\u001B[33m../dataset/open_cml.p\u001B[39m\u001B[33m'\u001B[39m) \u001B[38;5;66;03m#  read OpenCML dataset\u001B[39;00m\n",
+      "\u001B[31mAttributeError\u001B[39m: module 'pynncml' has no attribute 'read_open_cml_dataset'"
+     ]
+    }
+   ],
+   "execution_count": 2
   },
   {
    "cell_type": "markdown",
diff --git a/examples/wet_dry_classification.ipynb b/examples/wet_dry_classification.ipynb
diff --git a/pynncml/cml_methods/apis/xarray_processing/xarray_inference_engine.py b/pynncml/cml_methods/apis/xarray_processing/xarray_inference_engine.py
@@ -8,6 +8,15 @@
 from pynncml.datasets.xarray_processing import xarray2link
 from pynncml.multiple_cmls_methods import InferMultipleCMLs
 
+def create_dataset_with_coords_only(original_ds):
+    """
+    Creates a new, empty xarray Dataset with coordinates from a given dataset.
+    """
+    # Simply create a new Dataset, passing the coords from the original.
+    # The .coords property contains all the coordinate variables.
+    new_ds = xr.Dataset(coords=original_ds.coords)
+    return new_ds
+
 
 class XarrayInferenceEngine(nn.Module):
     def __init__(self,in_cml2rain_method:BaseCMLProcessingMethod,is_recurrent=True,is_attenuation=False, *args, **kwargs):
@@ -18,15 +27,15 @@ def __init__(self,in_cml2rain_method:BaseCMLProcessingMethod,is_recurrent=True,i
     def forward(self, x_xarray):
         link_set=xarray2link(x_xarray)
         results_data= self.inference_engine(link_set)
-
+        x_xarray_new=create_dataset_with_coords_only(x_xarray)
         new_var_dims = ('time', 'sublink_id', 'cml_id')
         new_var_coords = {
             'time': x_xarray.time,
             'sublink_id': x_xarray.sublink_id,
             'cml_id': x_xarray.cml_id
         }
         for rname in CMLResultsDataStructure.results_types_list():
-            x_xarray[rname] = xr.DataArray(
+            x_xarray_new[rname] = xr.DataArray(
             np.full((x_xarray.sizes['time'], x_xarray.sizes['sublink_id'], x_xarray.sizes['cml_id']), np.nan),
             coords=new_var_coords,
             dims=new_var_dims
@@ -43,6 +52,6 @@ def forward(self, x_xarray):
                 )
                 # Assign the new values to the dataset at the specified slice.
                 # This will create a new variable if it doesn't exist.
-                x_xarray[rname].loc[dict(sublink_id=link.sublink_id, cml_id=link.cml_id)] = new_values
+                x_xarray_new[rname].loc[dict(sublink_id=link.sublink_id, cml_id=link.cml_id)] = new_values
 
-        return x_xarray
+        return x_xarray_new
diff --git a/pynncml/datasets/link_data.py b/pynncml/datasets/link_data.py
@@ -392,7 +392,7 @@ def attenuation(self) -> torch.Tensor:
         :return attenuation: torch.Tensor
         """
         if self.has_tsl():
-            return torch.tensor(-(self.link_tsl - self.link_rsl)).reshape(1, -1).float()
+            return torch.tensor((self.link_tsl - self.link_rsl)).reshape(1, -1).float()
         else:
             return torch.tensor(-self.link_rsl).reshape(1, -1).float()
 
diff --git a/tests/test_data_structure.py b/tests/test_data_structure.py
@@ -74,7 +74,7 @@ def test_link_with_tsl(self):
         self.assertTrue(np.array_equal(l.time(), time.astype('datetime64[s]')))
         self.assertEqual(len(l.time()), TestDataStructure.n_samples)
         att = l.attenuation().numpy().flatten()
-        self.assertTrue(np.round(np.sum(att + tsl - rsl) * 100) == 0)
+        self.assertTrue(np.round(np.sum(att -tsl + rsl) * 100) == 0)
         l_min_max = l.create_min_max_link(10)
         self.assertTrue(len(l_min_max) == 10)
         self.assertEqual(len(l_min_max.attenuation().shape), 3)
diff --git a/tests/test_xarray_processing.py b/tests/test_xarray_processing.py
@@ -2,16 +2,16 @@
 
 import poligrain as plg
 
-from pynncml.apis.xarray_processing.wet_dry_methods import create_wet_dry_std
+from pynncml.cml_methods.apis.xarray_processing.wet_dry_methods import create_wet_dry_std
 
 
 class TestOpenCML(unittest.TestCase):
 
-    def test_poligrain_to_xarray(self):
+    def test_poligrain_to_xarray_openmrg(self):
         (ds_rad,
          ds_cmls,
          ds_gauges_municp,
          ds_gauge_smhi) = plg.example_data.load_openmrg(data_dir="example_data", subset="8d")
-        nn_base=create_wet_dry_std()
-        nn_base(ds_cmls)
+        nn_base=create_wet_dry_std(threshold=2.3,step=10)
+        ds_cmls_wet_dry=nn_base(ds_cmls)
 
