Updated Streamlit App with new ENR prediction model

Author: antoinetavant

pyproject.toml

@@ -35,7 +35,6 @@ dependencies = [
   "eccodes",
   "cfgrib",
   "requests",
-  "statsmodels",
   "geojson",
   "shapely",
   "tqdm",
@@ -80,7 +79,7 @@ build = "sphinx-build -b html doc/ doc/_build"
 [tool.hatch.envs.serve]
 extra-dependencies = [
   "streamlit",
-  "streamlit-analytics2",
+  # "streamlit-analytics2",
 ]
 
 [tool.hatch.envs.serve.scripts]

src/energy_forecast/dashboard/pages/2_Production.py

@@ -5,45 +5,15 @@
 """
 from energy_forecast.energy import ECO2MixDownloader
 from energy_forecast import ROOT_DIR
-from energy_forecast.meteo import get_region_sun, get_region_wind, memory
+from energy_forecast.meteo import ArpegeSimpleAPI, memory
+from energy_forecast.enr_production_model import ENRProductionModel
 import pandas as pd
 import streamlit as st
 import altair as alt
 alt.renderers.set_embed_options(time_format_locale="fr-FR", format_locale="fr-FR")
 
 @memory.cache
-def compute_data_pv_power(date: str)->pd.DataFrame:
-    """Estimate the power generated by solar panels.
-
-    Use the data from :func:`get_region_sun` and a linear model to estimate the power generated by solar panels.
-
-    Parameters
-    ----------
-    date : str
-        The date at which the power is estimated.
-
-    Returns
-    -------
-    pd.DataFrame
-        The power generated by solar panels for each hour of the day.
-        The columns are ["time", "pv_power"]
-    """
-    sun_data = get_region_sun(date)
-    sun_data.columns = [
-        c.replace(" ", "_").replace("'", "_").replace("-", "_").lower()
-        for c in sun_data.columns
-    ]
-    model_params = pd.read_csv(ROOT_DIR / "data/silver/sun_model_2_params.csv")
-    model_params.columns = ["region", "coef"]
-    model_params_ser = model_params.set_index("region").iloc[:, 0]
-    production = sun_data * model_params_ser
-    pv_power = production.sum(axis=1).to_frame() / 24
-    pv_power.reset_index(inplace=True)
-    pv_power.columns = ["time", "pv_power"]
-    return pv_power
-
-@memory.cache
-def compute_data_eolien(date)->pd.DataFrame:
+def compute_my_prodiction(date)->pd.DataFrame:
     """Estimate the power generated by eolien.
 
     Use the data from :func:`get_region_swind` and a linear model to estimate the power generated by wind turbines.
@@ -59,20 +29,12 @@ def compute_data_eolien(date)->pd.DataFrame:
         The power generated by solar panels for each hour of the day.
         The columns are ["time", "eolien_power"]
     """
-    wind_data = get_region_wind(date)
-    wind_data.columns = [
-        c.replace(" ", "_").replace("'", "_").replace("-", "_").lower()
-        for c in wind_data.columns
-    ]
-    model_params = pd.read_csv(ROOT_DIR / "data/silver/wind_model_2_params.csv")
-    model_params.columns = ["region", "coef"]
-    model_params_ser = model_params.set_index("region").iloc[:, 0]
-    production = wind_data * model_params_ser
-    eolien_power = production.sum(axis=1).to_frame() / 24
-    eolien_power.reset_index(inplace=True)
-    eolien_power.columns = ["time", "eolien_power"]
-    return eolien_power
-
+    wind_data = ArpegeSimpleAPI(date).departement_wind()
+    sun_data = ArpegeSimpleAPI(date).departement_sun()
+    my_model = ENRProductionModel.load(filename="model_departements.pkl")
+    
+    predictions = my_model.predict(sun_data, wind_data)
+    return predictions
 
 @memory.cache
 def compute_energy(date:str):
@@ -101,8 +63,8 @@ def compute_energy(date:str):
         - ``"Eolien Prediction"``: the wind power production prediction (from linear model using weather date)
 
     """
-    pv_power = compute_data_pv_power(date)
-    eolien_power = compute_data_eolien(date)
+    predictions = compute_my_prodiction(date)
+    
     current_year = pd.Timestamp(date).year
     r = ECO2MixDownloader(year=current_year)
     r.download()
@@ -116,9 +78,9 @@ def compute_energy(date:str):
     energy.index = energy.index.tz_localize("Europe/Paris").tz_convert("UTC").tz_localize(None)
     energy = pd.concat([
         energy,
-        pv_power.set_index("time").rename(columns={"pv_power": "PV Prediction"}),
-        eolien_power.set_index("time").rename(columns={"eolien_power": "Eolien Prediction"})
-    ], axis=1).reset_index()
+        predictions.rename(columns={"sun": "PV Prediction",
+                                    "wind": "Eolien Prediction"})
+    ], axis=1).reset_index(names="time")
 
     return energy
 
@@ -129,6 +91,7 @@ def compute_energy(date:str):
     # date = pd.Timestamp.now().strftime("%Y-%m-%d")
     energy = compute_energy(date_input)
     # keep only full hours, needed to plot the data without empty spaces
+    print(energy)
     energy = energy[energy["time"].dt.minute == 0]
 
     st.title("Prévision de production ENR")
@@ -162,4 +125,4 @@ def compute_energy(date:str):
     )
     st.altair_chart(sun_chart,
                     use_container_width=True
-                )
+                )