AutoML_LCZ_classfier.py

import pandas as pd
import numpy as np
from flaml import AutoML
from sklearn.model_selection import StratifiedShuffleSplit
from sklearn.metrics import accuracy_score
from sklearn.utils import resample
import joblib
import pickle
import matplotlib.pyplot as plt

# 保存 splits 到 pickle 文件
with open('your_Manchester_TA_splits.pkl', 'rb') as f:
    splits = pickle.load(f)

with open('your_Manchester_TA_features.pkl', 'rb') as f:
    polygon_features = pickle.load(f)

for i, (train_gdf, test_gdf) in enumerate(splits):
    # 合并训练和测试数据
    train_data = pd.concat([polygon_features[idx] for idx in train_gdf.index], ignore_index=True)
    test_data = pd.concat([polygon_features[idx] for idx in test_gdf.index], ignore_index=True)
    training_all = pd.concat([train_data, test_data], axis=0)
    # 提取特征和标签
    X_train = training_all.drop(['Name'], axis=1)
    y_train = training_all['Name'].astype('int32')
    break
   # 定义 AutoML
automl = AutoML()
settings = {
    "time_budget": 3600,  # 时间预算
    "task": "classification",  # 任务类型
    "metric": "accuracy",  # 优化目标
    "log_file_name": f"/home/jiyang/P2_LCZ/log/automl_4lcz_new_training_20250130.log"
}
# 训练模型
automl.fit(X_train=X_train.values, y_train=y_train.values, **settings)
# 保存最佳模型
model_path = f"/home/jiyang/P2_LCZ/ml_outputs/best_newRS4LCZ_Manchester_20250130.pkl"
# Save the model
with open(f"{model_path}", "wb") as f:
    pickle.dump(automl, f, pickle.HIGHEST_PROTOCOL)
print(f"Model for Manchester LCZ mapping saved as {model_path}")