diff --git a/notebooks/plotting.ipynb b/notebooks/plotting.ipynb
index 4825ab4bbbc442da9a562bb283d96d1b42153c2d..66914d34eff2c98015a51ff0827583d1e5ba7283 100644
--- a/notebooks/plotting.ipynb
+++ b/notebooks/plotting.ipynb
@@ -2,7 +2,7 @@
"cells": [
{
"cell_type": "code",
- "execution_count": null,
+ "execution_count": 1,
"id": "e0d6a2e3",
"metadata": {},
"outputs": [],
@@ -41,14 +41,222 @@
" accelerator = 'cpu'"
]
},
+ {
+ "cell_type": "markdown",
+ "id": "b1aec6bf",
+ "metadata": {},
+ "source": [
+ "# Neuer Versuch"
+ ]
+ },
+ {
+ "cell_type": "code",
+ "execution_count": 153,
+ "id": "13b1bcc5",
+ "metadata": {},
+ "outputs": [],
+ "source": [
+ "from utils.helpers import load_model\n",
+ "from torch.utils.data import DataLoader\n",
+ "from data_tools.datasets import TimeSeriesDataSet\n",
+ "from plotly.subplots import make_subplots\n",
+ "\n",
+ "model_dir = Path('../KIWaVo/models/lfu/sehmsdorf/lightning_logs/version_10/')\n",
+ "data_file = Path('../KIWaVo/data/input/MergedSehmsdorfAll.csv')"
+ ]
+ },
+ {
+ "cell_type": "code",
+ "execution_count": 166,
+ "id": "c53eb9aa",
+ "metadata": {},
+ "outputs": [],
+ "source": [
+ "def load_df_pred(data_file : Path,model):\n",
+ " df = pd.read_csv(data_file, index_col=0, parse_dates=True)\n",
+ " if not isinstance(df.index,pd.core.indexes.datetimes.DatetimeIndex):\n",
+ " df = pd.read_csv(data_file,index_col=0,parse_dates=True,date_format='%d.%m.%Y %H:00')\n",
+ "\n",
+ " df = ut.fill_missing_values(df)\n",
+ "\n",
+ " if model.hparams.differencing == 1:\n",
+ " df[\"d1\"] = df[df.columns[model.hparams.gauge_idx]].diff()\n",
+ "\n",
+ " if model.hparams.embed_time:\n",
+ " hours = df.index.hour\n",
+ " hour_rad = torch.tensor(2*torch.pi*hours/24)\n",
+ " df[\"h_sin\"] = torch.sin(hour_rad)\n",
+ " df[\"h_cos\"] = torch.cos(hour_rad)\n",
+ "\n",
+ " days = df.index.day_of_year + hours / 24\n",
+ " days_rad = torch.tensor(2*torch.pi*days/365)\n",
+ " df[\"d_sin\"] = torch.sin(days_rad)\n",
+ " df[\"d_cos\"] = torch.cos(days_rad)\n",
+ " columns = ['h_sin', 'h_cos', 'd_sin', 'd_cos'] + [col for col in df.columns if col not in ['h_sin', 'h_cos', 'd_sin', 'd_cos']]\n",
+ " df = df[columns]\n",
+ "\n",
+ " return df[1:]\n",
+ "\n",
+ "def load_data_pred(data_file,model):\n",
+ " df = load_df_pred(data_file,model)\n",
+ " df_scaled = pd.DataFrame(model.scaler.transform(df),index=df.index,columns=df.columns)\n",
+ " x = torch.Tensor(df_scaled.values) #[start:end]\n",
+ " y = torch.Tensor(df_scaled[df.columns[model.hparams.target_idx]].values)#[start:end]]\n",
+ "\n",
+ " dataset = TimeSeriesDataSet(x,y,model.in_size,model.out_size)\n",
+ "\n",
+ " y_true = df[df.columns[model.hparams.gauge_idx]][model.in_size-1:-model.out_size-1]\n",
+ "\n",
+ " data_loader = DataLoader(dataset, batch_size=256,num_workers=8)\n",
+ "\n",
+ " return df, df_scaled, data_loader,y_true\n",
+ "\n",
+ "def plot(y_pred,start=None,end=None,cols=None):\n",
+ " if start is None:\n",
+ " start = y_pred.index[0]\n",
+ " elif isinstance(start,str):\n",
+ " start = pd.to_datetime(start)\n",
+ " if end is None:\n",
+ " end = y_pred.index[-1]\n",
+ " elif isinstance(end,str):\n",
+ " end = pd.to_datetime(end)\n",
+ " if cols is None:\n",
+ " cols = [1,12,24,48]\n",
+ "\n",
+ " y_pred_sel = y_pred[start:end]\n",
+ "\n",
+ " fig = go.Figure()\n",
+ " fig.add_trace(go.Scatter(x=y_pred_sel.index,y=y_pred_sel[0],name='True',line=dict(width=3)))\n",
+ "\n",
+ " for i in cols:\n",
+ " fig.add_trace(go.Scatter(x=y_pred_sel.index[i:], y=y_pred_sel[i][:-i], mode='lines', name=f\"Pred {i}\"))\n",
+ " return fig\n",
+ "\n"
+ ]
+ },
+ {
+ "cell_type": "code",
+ "execution_count": null,
+ "id": "43fc20f5",
+ "metadata": {},
+ "outputs": [],
+ "source": []
+ },
{
"cell_type": "code",
"execution_count": null,
- "id": "e4ac7073",
+ "id": "044c3bbb",
+ "metadata": {},
+ "outputs": [],
+ "source": [
+ "model = load_model(model_dir)\n",
+ "df, df_scaled, data_loader,y_true= load_data_pred(data_file,model)\n",
+ "y_pred = hp.get_pred(model,data_loader,y_true)"
+ ]
+ },
+ {
+ "cell_type": "code",
+ "execution_count": 143,
+ "id": "54c6c264",
"metadata": {},
"outputs": [],
"source": []
},
+ {
+ "cell_type": "code",
+ "execution_count": 169,
+ "id": "ab939e21",
+ "metadata": {},
+ "outputs": [],
+ "source": [
+ "\n",
+ "def plot_cont(y_pred,start=None,end=None,hours=None):\n",
+ " if start is None:\n",
+ " start = y_pred.index[0]\n",
+ " elif isinstance(start,str):\n",
+ " start = pd.to_datetime(start)\n",
+ " if end is None:\n",
+ " end = y_pred.index[-1]\n",
+ " elif isinstance(end,str):\n",
+ " end = pd.to_datetime(end)\n",
+ " if hours is None:\n",
+ " hours = [0]\n",
+ "\n",
+ " y_pred_sel = y_pred[start:end]\n",
+ " y_pred_sel2 = y_pred_sel[y_pred_sel.index.hour.isin(hours)]\n",
+ "\n",
+ " fig = go.Figure()\n",
+ " fig.add_trace(go.Scatter(x=y_pred_sel.index,y=y_pred_sel[0],name='True',line=dict(width=3)))\n",
+ "\n",
+ " # Create figure with secondary y-axis\n",
+ " fig = make_subplots(specs=[[{\"secondary_y\": True}]])\n",
+ "\n",
+ " fig.add_trace(go.Scatter(x=y_pred_sel.index,y=y_pred_sel[0],name='True',line=dict(width=5)))\n",
+ " for idx, row in y_pred_sel2.iterrows():\n",
+ "\n",
+ " x = pd.date_range(idx,periods=48+1,freq='h')[1:]\n",
+ " y = row[1:]\n",
+ " fig.add_trace(go.Scatter(x=x,y=y,name=idx.strftime('%Y-%m-%d %h')))\n",
+ "\n",
+ "\n",
+ " return fig\n"
+ ]
+ },
+ {
+ "cell_type": "code",
+ "execution_count": null,
+ "id": "bf4f39fa",
+ "metadata": {},
+ "outputs": [],
+ "source": [
+ "start=pd.to_datetime('2024-01-01')\n",
+ "end=pd.to_datetime('2024-06-02')\n",
+ "fig = plot_cont(y_pred,start,end,hours=[0,6,12,18])\n",
+ "\n",
+ "fig.add_trace(go.Scatter(x=df[start:end].index,y=df[\"NSehmsdorf_mm\"][start:end],name='Rain'),secondary_y=True)\n",
+ "fig"
+ ]
+ },
+ {
+ "cell_type": "code",
+ "execution_count": null,
+ "id": "9d86847e",
+ "metadata": {},
+ "outputs": [],
+ "source": [
+ "plot(y_pred,start='2024-01-01',end='2024-06-02',cols=[1,12,24,48])"
+ ]
+ },
+ {
+ "cell_type": "code",
+ "execution_count": null,
+ "id": "e204fc12",
+ "metadata": {},
+ "outputs": [],
+ "source": [
+ "start=pd.to_datetime('2024-01-01')\n",
+ "end=pd.to_datetime('2024-06-02')\n",
+ "\n",
+ "y_pred_sel = y_pred[start:end]\n",
+ "\n",
+ "fig = go.Figure()\n",
+ "\n",
+ "fig.add_trace(go.Scatter(x=y_pred_sel.index,y=y_pred_sel[0],name='True',line=dict(width=3)))\n",
+ "\n",
+ "for i in [1,24,48]:\n",
+ " fig.add_trace(go.Scatter(x=y_pred_sel.index[i:],y=y_pred_sel[i][:-i],name=f'Pred {i}'))\n",
+ "\n",
+ "fig"
+ ]
+ },
+ {
+ "cell_type": "markdown",
+ "id": "11ae694c",
+ "metadata": {},
+ "source": [
+ "# Alter Versuch"
+ ]
+ },
{
"cell_type": "code",
"execution_count": null,
@@ -1573,6 +1781,34 @@
"fig.update_layout(xaxis_title=\"Zeit\",yaxis_title=\"Wasserstand [cm]\")\n",
"fig.write_html(\"pötrau.html\")"
]
+ },
+ {
+ "cell_type": "code",
+ "execution_count": 1,
+ "id": "f1d5c345",
+ "metadata": {},
+ "outputs": [],
+ "source": [
+ "temp = {\"a\":1,\"b\":2}"
+ ]
+ },
+ {
+ "cell_type": "code",
+ "execution_count": 3,
+ "id": "892f855c",
+ "metadata": {},
+ "outputs": [],
+ "source": [
+ "temp.get(\"c\")"
+ ]
+ },
+ {
+ "cell_type": "code",
+ "execution_count": null,
+ "id": "d96cf6a1",
+ "metadata": {},
+ "outputs": [],
+ "source": []
}
],
"metadata": {