Autogluon傻瓜式深度学习

提示：这里简述项目相关背景：

安装教程可参考：
教程：
https://blog.csdn.net/lichangzhen2008/article/details/117673972

例如：可使用.csv文件输入：第一行为各种数据的标签，通过下面代码指定target标签
‘regression’: 回归问题
‘classfication’: 分类问题

predictor = tabular.TabularPredictor(label='Label_mL',problem_type='regression').fit(train_data=train_x)

完整示例代码

import autogluon as ag
from autogluon import tabular
import pandas as pd
from sklearn.model_selection import train_test_split
import math
from sklearn import metrics

df = tabular.TabularDataset(r'C:\Users\96552\Desktop\autogluon\New_50gap.csv')
X=df
y=df['Label_mL']

train_x,test_x,train_y,test_y=train_test_split(X,y,test_size=0.25,random_state=0)

from sklearn.model_selection import train_test_split



predictor = tabular.TabularPredictor(label='Label_mL',problem_type='regression').fit(train_data=train_x)
test_data = tabular.TabularDataset(test_x)
leaderboard = predictor.leaderboard(test_data)

test_data_t=test_x.drop(labels=['Label_mL'],axis=1)
pred_y=predictor.predict(test_data,model='LightGBM')

print("MSE (Test): ", metrics.mean_squared_error(test_y, pred_y))
# Compute the RMSE
print("RMSE (Test): ", math.sqrt(metrics.mean_squared_error(test_y, pred_y)))
# Compute the MAE
print("MAE (Test): ", metrics.mean_absolute_error(test_y, pred_y))

print(test_y-pred_y)

训练中：

训练结果：会提供训练集准确率、测试集准确率、训练时间、排名等。
回归问题问题中可能出现RMSE为负数的情况，不是错误，经验证取绝对值即可。