change to class

data/preprocessing/single_prediction_features.py

@@ -0,0 +1,84 @@
+import pandas as pd
+import openmeteo_requests
+import requests_cache
+from datetime import datetime, date, timedelta
+from retry_requests import retry
+
+from .prepare_time_features import prepare_time_features
+
+# TODO get these from metadata
+feature_columns = ['ferien', 'feiertag', 'covid_19', 'olma_offa', 'temperature_2m_max',
+                   'temperature_2m_min', 'rain_sum', 'snowfall_sum', 'sin_minute',
+                   'cos_minute', 'sin_hour', 'cos_hour', 'sin_weekday', 'cos_weekday',
+                   'sin_day', 'cos_day', 'sin_month', 'cos_month']
+weather_api_url = "https://api.open-meteo.com/v1/forecast"  # URL API
+
+
+class SinglePredictionFeatures:
+    def __init__(self, raw_features_path):
+        # Import Features
+        self.calendar_features = pd.read_csv(raw_features_path, sep=",")
+        self.calendar_features['date'] = pd.to_datetime(self.calendar_features['date'],
+                                                        format='%Y-%m-%d')  # Extract Date
+
+        # Get Weather-Data
+        # Set up Client and define Params
+        cache_session = requests_cache.CachedSession('.cache', expire_after=3600)
+        retry_session = retry(cache_session, retries=5, backoff_factor=0.2)
+        self.openmeteo = openmeteo_requests.Client(session=retry_session)
+
+    def get_weather_forecast(self, params):
+        responses = self.openmeteo.weather_api(weather_api_url, params=params)
+        response = responses[0]
+
+        daily = response.Daily()
+        daily_temperature_2m_max = daily.Variables(0).ValuesAsNumpy()
+        daily_temperature_2m_min = daily.Variables(1).ValuesAsNumpy()
+        daily_rain_sum = daily.Variables(2).ValuesAsNumpy()
+        daily_snowfall_sum = daily.Variables(3).ValuesAsNumpy()
+
+        daily_data = {"date": pd.date_range(
+            start=pd.to_datetime(daily.Time(), unit="s"),
+            end=pd.to_datetime(daily.TimeEnd(), unit="s"),
+            freq=pd.Timedelta(seconds=daily.Interval()),
+            inclusive="left"
+        ), "temperature_2m_max": daily_temperature_2m_max, "temperature_2m_min": daily_temperature_2m_min,
+            "rain_sum": daily_rain_sum, "snowfall_sum": daily_snowfall_sum}
+
+        daily_dataframe = pd.DataFrame(data=daily_data)
+        return daily_dataframe
+
+    def build_dataframe(self, input_date):
+        timestamp = datetime.strptime(input_date, '%Y-%m-%d %H:%M')
+
+        params = {
+            "latitude": 47.4239,
+            "longitude": 9.3748,
+            "daily": ["temperature_2m_max", "temperature_2m_min", "rain_sum", "snowfall_sum"],
+            "start_date": timestamp.strftime("%Y-%m-%d"),
+            "end_date": timestamp.strftime("%Y-%m-%d")
+        }
+
+        # Get Data
+        df_weather = self.get_weather_forecast(params)
+
+        # Merge Weather with other Features
+        df = pd.merge(df_weather, self.calendar_features, on="date", how="left")
+        df["datetime"] = timestamp
+
+        # Add time-features
+        df = prepare_time_features(df)
+
+        df_filtered = df[feature_columns]
+
+        return df_filtered, len(feature_columns)
+
+
+if __name__ == "__main__":
+    date_today = date.today()
+    date_tomorrow = date_today + timedelta(days=1)
+    single_prediction_features = SinglePredictionFeatures("raw_features_2024.csv")
+    df_demo, features_length = single_prediction_features.build_dataframe(date_tomorrow.strftime("%Y-%m-%d %H:%M"))
+    print(df_demo.head())
+    print(df_demo.columns)
+    print(features_length)

deploy/predict.py

@@ -1,20 +1,21 @@
 import torch
 from torch.utils.data import TensorDataset, DataLoader
-from data.preprocessing.get_features_for_prediction import build_dataframe
+from data.preprocessing.single_prediction_features import SinglePredictionFeatures
 
 batch_size = 1  # Required for model input
 parking_data_labels = ["P24", "P44", "P42", "P33", "P23", "P25", "P21", "P31", "P53", "P32", "P22", "P52", "P51",
                        "P43"]  # TODO get these from metadata file
+raw_features_path = "data/preprocessing/raw_features_2024.csv"
 
 
 class Predict:
     def __init__(self, model_path):
+        self.single_prediction_features = SinglePredictionFeatures(raw_features_path)
         self.device = torch.device("cuda" if torch.cuda.is_available() else "cpu")
         self.model = torch.jit.load(model_path, map_location=self.device)
         self.model.eval()
 
-    @staticmethod
-    def build_dataset(df):
+    def build_dataset(self, df):
         features = torch.Tensor(df.values)
 
         dataset = TensorDataset(features)
@@ -23,13 +24,18 @@ def build_dataset(df):
 
     def predict_with_model(self, dataloader, features_length):
         data = next(iter(dataloader))[0]
-        data = data.view([batch_size, -1, features_length]).to(device)
+        data = data.view([batch_size, -1, features_length]).to(self.device)
         return self.model(data).cpu()
 
     def predict_for_date(self, date):
-        features_df, features_length = build_dataframe(date)
+        features_df, features_length = self.single_prediction_features.build_dataframe(date)
         dataloader = self.build_dataset(features_df)
 
         output = self.predict_with_model(dataloader, features_length)
 
         return [dict(zip(parking_data_labels, row)) for row in output.tolist()]
+
+
+if __name__ == "__main__":
+    predict = Predict("model_scripted.pt")
+    print(predict.predict_for_date("2023-10-09 00:00"))