import os
import tensorflow as tf
from keras.models import load_model
from tensorflow.keras.losses import MeanSquaredError
import pandas as pd
from sklearn.preprocessing import OneHotEncoder, StandardScaler

# Disable GPU if not needed
os.environ["CUDA_VISIBLE_DEVICES"] = "-1"  # Disable GPU for CPU-only execution

# Function to load the model and set the loss function explicitly
def load_and_compile_model(model_path):
    model = load_model(model_path, custom_objects={'MeanSquaredError': MeanSquaredError})
    model.compile(optimizer='adam', loss=MeanSquaredError())
    return model

# Load the model
loaded_model = load_and_compile_model('solar_irradiance_model.h5')

# Load the dataset
data = pd.read_csv('Solar_Irradiance.csv')

# Example data preparation (adjust as needed based on your dataset)
data['Latitude'] = data['Latitude'].str.rstrip('°').astype(float)
data['Longitude'] = data['Longitude'].str.rstrip('°').astype(float)

# Define feature columns
features = data[['Month', 'Hour', 'Latitude', 'Longitude', 'Panel_Capacity(W)', 'Panel_Efficiency', 'Wind_Speed(km/h)', 'Cloud_Cover(%)', 'temperature (°f)']]

# One-hot encoding for categorical variables
encoder = OneHotEncoder(sparse_output=False, categories='auto')
categorical_features = features[['Month', 'Hour']]
encoder.fit(categorical_features)

# Scaling numerical features
scaler = StandardScaler()
numerical_features = features[['Latitude', 'Longitude', 'Panel_Capacity(W)', 'Panel_Efficiency', 'Wind_Speed(km/h)', 'Cloud_Cover(%)', 'temperature (°f)']]
scaler.fit(numerical_features)

# Prepare the data by transforming categorical and numerical features
encoded_categorical_features = encoder.transform(categorical_features)
scaled_numerical_features = scaler.transform(numerical_features)

# Combine the transformed features
processed_features = pd.DataFrame(encoded_categorical_features, columns=encoder.get_feature_names_out())
processed_features[numerical_features.columns] = scaled_numerical_features

# Assuming you have a target column, for example 'Solar_Irradiance', adjust as needed
target = data['Solar_Irradiance']  # Make sure to replace this with your actual target column name

# Make predictions using the loaded model
predictions = loaded_model.predict(processed_features)

# Print or save the predictions
print(predictions)