22亚太杯数学建模竞赛C题全球气温预测代码思路

提示：文章写完后，目录可以自动生成，如何生成可参考右边的帮助文档

前言

22亚太杯数学建模竞赛C题全球气温预测代码思路

一、问题描述？

竞赛提供的数据集总共有239177条，包含了从1820年到2013年的全球100个城市的气温数据，要进行温度预测，首先要将数据处理一下，时空数据预测可用的机器学习方法很多，这里提供的例子为LightGBM。

二、使用步骤

1.引入库

代码如下：

import pandas as pd

import numpy as np

import lightgbm as lgb

from datetime import datetime

from sklearn import preprocessing

from sklearn import metrics

import pickle

from sklearn.metrics import roc_auc_score

from sklearn.model_selection import train_test_split

from sklearn.metrics import accuracy_score

from sklearn import metrics

import time

import random

import math

import matplotlib.pyplot as plt

import warnings

import xgboost as xgb

from sklearn.model_selection import StratifiedKFold, KFold

from sklearn.metrics import cohen_kappa_score

import seaborn as sns

from datetime import datetime

from sklearn import preprocessing

from sklearn.metrics import roc_auc_score

from sklearn.model_selection import train_test_split

2.读入数据

代码如下：

df=pd.read_csv('2022_APMCM_C_Data.csv')
df

3.数据处理

代码如下：

df['dt'] = pd.to_datetime(df['dt'])
#df['年月日'] = df['dt'].apply(lambda x: x.strftime('%Y%m%d'))
df['年']=df['dt'].dt.year 
df['月']=df['dt'].dt.month
df['日']=df['dt'].dt.day
df.drop(['dt'],axis=1,inplace=True)
df
warnings.simplefilter(action='ignore', category=FutureWarning)
colum = ['Latitude', 'Longitude']
data=df
lbl = preprocessing.LabelEncoder()
for strflag in colum:
    data[strflag] = lbl.fit_transform(data[strflag].astype(str))
final_labels = data.columns.tolist()[1:6]
#data.replace('?',np.nan,inplace=True)
#data.replace('',np.nan,inplace=True)
#data.replace(-np.inf, np.nan,inplace=True)
#data.replace(np.inf, np.nan,inplace=True)
data=data.dropna(axis=0)
data

4.划分训练集和测试集

代码如下：

train_name=data.columns[1:6]
train=data[train_name]
label=data['AverageTemperature']

train_x, TEST_X, train_y, TEST_y = train_test_split(
    train,
    label,
    test_size=0.3,
    random_state=1,
)

train_y=pd.DataFrame(train_y)
train_y.columns=['AverageTemperature']
TEST_y=pd.DataFrame(TEST_y)
TEST_y.columns=['AverageTemperature']

train_x.reset_index(inplace=True,drop=True)
train_y.reset_index(inplace=True,drop=True)
TEST_X.reset_index(inplace=True,drop=True)
TEST_y.reset_index(inplace=True,drop=True)

5.训练

代码如下：

## 参数
params = {
    'boosting_type': 'gbdt',
    'objective': 'regression',
    'metric': 'rmse',
    'bagging_freq':2,
    'num_leaves': 511,
    'max_depth': 9,
    'device': 'cpu',
    'min_data_in_leaf': 220,
    'learning_rate': 0.01,
    'feature_fraction': 0.8,
    'bagging_fraction': 0.9,
    'lambda_l1': 0.3,
    'lambda_l2': 0.5,
    'min_gain_to_split': 0.2,
    'is_unbalance': True,
    'subsample':0.9,
    'colsample_bytree': 0.7,
    'colsample_bylevel': 0.7,
    'eta': 0.04,
}
##训练

from lightgbm import LGBMRegressor
model1 = LGBMRegressor(learning_rate=0.2,
                      num_leaves=255,
                      bagging_freq=2,
                      metric='rmse',
                      max_depth=8,
                      min_data_in_leaf=220,
                      feature_fraction=0.9,
                      bagging_fraction=0.9,
                      lambda_l1=0.3,
                      lambda_l2=0.5,
                      min_gain_to_split=0.2,
                      subsample=0.9,
                      colsample_bytree=0.7,
                      colsample_bylevel=0.7
                     )
model1.fit(train_x,train_y)
y_predict=model1.predict(TEST_X)
y_predict
y_predict=pd.DataFrame(y_predict)
y_predict.columns=['AverageTemperature1']
##测试
test_err=pd.concat([y_predict,TEST_y],axis=1)
test_err1=test_err.dropna(axis=0)
test_err1

---