def do_setup():
import aml_utils
####################################################################
# Setup- train model, create direct db connections, set global constants, etc.
#####################################################################
# TODO: create another model that uses a different algo (logisticRegression, perhaps), so we can have tests for our logic when using multiple models but each predicting off the same features
ml_predictor_titanic, df_titanic_test = aml_utils.train_basic_binary_classifier()
file_name = '_test_suite_saved_pipeline.dill'
ml_predictor_titanic.save(file_name)
ml_predictor_titanic = load_ml_model(file_name)
os.remove(file_name)
# row_ids = [i for i in range(df_titanic_test.shape[0])]
# df_titanic_test['row_id'] = row_ids
persistent_db_config = {
'db': '__concordia_test_env'
, 'host': 'localhost'
, 'port': 27017
}
in_memory_db_config = {
'db': 8
, 'host': 'localhost'
, 'port': 6379
}
host = in_memory_db_config['host']
port = in_memory_db_config['port']
db = in_memory_db_config['db']
rdb = redis.StrictRedis(host=host, port=port, db=db)
host = persistent_db_config['host']
port = persistent_db_config['port']
db = persistent_db_config['db']
client = MongoClient(host=host, port=port)
mdb = client[db]
concord = load_concordia(persistent_db_config=persistent_db_config)
existing_training_rows, _, _ = concord._get_training_data_and_predictions(model_id)
len_existing_training_rows = existing_training_rows.shape[0]
existing_live_rows = concord.retrieve_from_persistent_db(val_type='live_features', row_id=None, model_id=model_id)
len_existing_live_rows = len(existing_live_rows)
return ml_predictor_titanic, df_titanic_test, concord, rdb, mdb, len_existing_training_rows, len_existing_live_rows
"tradeTypeId": 'categorical',
# 'bedrooms': 'categorical',
# 'year': 'categorical',
# 'month': 'categorical',
}
# print(column_description1)
# # 合并两个字典
# column_descriptions = dict(column_description1, **column_description2)
#
# ml_predictor = Predictor(type_of_estimator='Regressor', column_descriptions=column_descriptions)
#
# ml_predictor.train(df_train,model_names='XGBRegressor')
# ml_predictor.save('model_auto_ml_9.h5')
ml_predictor = load_ml_model('auto_ml_new.h5')
# 预测预测数据
x = ml_predictor.predict(df_test)
x_dataframe = pd.DataFrame(x,columns=['predictions'])
merge_data = pd.concat((origin_data,x_dataframe),axis=1)
merge_data_df = pd.DataFrame(merge_data)
merge_data_df.to_csv('./merge_data_bak/merge_data_auto_ml.csv',index=False)
print(x_dataframe.describe())
print(df_test_label.describe())
print(mean_absolute_error(df_test_label,x))
compute_ratio(merge_data_df)
# compute_ratio2(merge_data_df)
print(len(data_20) / len(data))
print(len(data_30) / len(data))
print(len(data_more) / len(data))
print(mean_absolute_error(data[test_column], data['daysOnMarket']))
if __name__ == '__main__':
# 导入数据
data = pd.read_csv('./input/treb_toronto_9.csv')
data = preprocess_data(data)
origin_data = data.reset_index(drop=True)
data_label = data['daysOnMarket']
data_prediction = data.drop(columns=['daysOnMarket'])
# 导入模型
model = load_ml_model('./some_fine_model/model_auto_ml_9.h5')
# 预测：
prediction_result = model.predict(data_prediction)
x = prediction_result
x_dataframe = pd.DataFrame(x, columns=['predictions'])
merge_data = pd.concat((origin_data, x_dataframe), axis=1)
merge_data_df = pd.DataFrame(merge_data)
merge_data_df.to_csv(
'./intermediate_generated_file/load_model_test_merge_prediction_result_data.csv',
index=False)
print(x_dataframe.describe())
print(data_label.describe())
print(mean_absolute_error(data_label, x))
compute_ratio(merge_data_df)
def my_predict_auto_ml(self, data):
model = load_ml_model(model_auto_ml_path)
result = model.predict(data)
return result