版权声明：不要转载复制当原创就好了，指明下参考地址或者书目，大家一起学习进步。 https://blog.csdn.net/Monk_donot_know/article/details/86088247
笔记说明：超链接加不进来！！！一加进来就卡死，真是崩溃！openpyxl这块不是我写的，是从一个人的博客上摘下来的。
就是下面这个。
https://blog.csdn.net/weixin_43094965/article/details/82226263
好神奇，上述这个链接居然是手打出来的，居然可以显示，复制粘贴过来就会卡死！
后面的案例部分是我工作写的，反正你们没数据，我就大胆摆出来了，欢迎指正。
dataframe
#文件(示例，data数据框并不存在)
import pandas as pd
a=pd.DataFrame(data)
a.head(6)
a.describe()
a.T ###转置
a.sort_index(axis=1,asceding=True)
a['x']
a[0:3]
a.loc[:,['a','b']] ##通过标签来索引
a.iloc[1:3,1:6] ###通过位置来索引
a.iloc[1:2]
a.iloc[[0,2,3],[1,2,7]] ##选取任意的行和列
a.loc[:,1:3]=7 ##直接赋值或者改变内容
read_excel()
读取数据之后原始数据第一行会变为标签，第二行变为第一行
pd.read_excel(io, sheet_name=0, header=0, names=None, index_col=None,
usecols=None, squeeze=False,dtype=None, engine=None,
converters=None, true_values=None, false_values=None,
skiprows=None, nrows=None, na_values=None,
parse_dates=False,
date_parser=None, thousands=None, comment=None, skipfooter=0,
convert_float=True, **kwds)
"""
io:excel的存储路径
sheet_name:要读取的工作表名称
header:用哪一行作列名
names:自定义最终的列名
index_col:用作索引的列
usecols:需要读取哪些列
converters:强行规定列数据类型
skiprows:跳过特定行
nrows:需要读取的行数
skipfooter:跳过末尾n行
"""
io=r'C:\Users\Administrator\Documents\WeChat Files\data'
sheet_name='中英文名都可以或者Sheet1' #S要大写,默认0(即Sheet1)
names=['a','b','c','d'] ##命名列
usecols=[0,1,2]
usecols='A:C,E' ##包括C
converters= {'排名':str,'场次':int} ##默认全部是整型
openpyxl
## 创建xlsx
from openpyxl import Workbook
wb = Workbook() ## 实例化，对excel数据表的操作在wb上
ws = wb.active ##激活sheet
## 读取xlsx
load_workbook(filename,read_only=False, use_iterators=False,
keep_vba=False, guess_types=False,
data_only=False)
from openpyxl import load_workbook
wb = load_workbook('文件名.xlsx') ## 默认只读第一个sheet
##一个例子
def read_excel_xlsx(path, sheet_name):
workbook = openpyxl.load_workbook(path)
# sheet = wb.get_sheet_by_name(sheet_name)这种方式已经弃用，不建议使用
sheet = workbook[sheet_name]
for row in sheet.rows:
for cell in row:
print(cell.value, "\t", end="")
print()
## 创建新的sheet
help(openpyxl.workbook.Workbook.create_sheet)
create_sheet(self, title=None, index=None)
##一个例子
def save(data1,K,path):
f = load_workbook(path) # 创建工作簿
sheet1 = f.create_sheet(title=para_phase, index=0) # 创建sheet
L = len(data1)
for j in range(0,L):
sheet1.cell(j+1,1).value = float(data1[j])
f.save(path) ## ！！！！重要！！！必须保存！！！！
一个案例，做随机森林，并将一些列指标计算出来储存在excel中。
import pandas as pd
import numpy as np
import pickle
from sklearn.ensemble import RandomForestRegressor
from sklearn.metrics import confusion_matrix
from sklearn.model_selection import KFold
from sklearn.metrics import roc_auc_score
import xlwt
import sys
import datetime
from openpyxl import load_workbook
from imblearn.over_sampling import SMOTE
global para_phase
para_phase = "phase11" #phase1，phase2，phase3。。。。。。。。。。 其中phase1 为未删特征版本
global balance
balance = "TRUE" #TRUE/FALSE
def save(data1,data2,data3,data4,data5,data6,data7,data8,path):
f = load_workbook(path) # 创建工作簿
sheet1 = f.create_sheet(title='model_%s'%(para_phase), index=0) # 创建sheet
L = len(data1) # h为行数，l为列数
for j in range(0,L):
sheet1['A1'].value = 'TN'
sheet1['A2'].value = 'FP'
sheet1['A3'].value = 'FN'
sheet1['A4'].value = 'TP'
sheet1['A5'].value = 'accuary rate'
sheet1['A6'].value = 'OOB Score'
sheet1['A7'].value = 'KS'
sheet1['A8'].value = 'AUC'
sheet1.cell(1,j+2).value = int(data1[j])
sheet1.cell(2,j+2).value = int(data2[j])
sheet1.cell(3,j+2).value = int(data3[j])
sheet1.cell(4,j+2).value = int(data4[j])
sheet1.cell(5,j+2).value = float(data5[j])
sheet1.cell(6,j+2).value = float(data6[j])
sheet1.cell(7,j+2).value = float(data7[j])
sheet1.cell(8,j+2).value = float(data8[j])
f.save(path)
if __name__ == '__main__':
all_data = pd.read_excel('TRAIN_all_%s.xlsx'%(para_phase),'Sheet1')
train_array = all_data.values
train_row = np.size(train_array,0)
train_col = np.size(train_array,1)
train_data =train_array[0:train_row,3:train_col]
train_re = train_array[0:train_row,2]
# train_re = train_re.astype(numpy.int)
kf = KFold(n_splits=10,shuffle=True)
predict_score = []
oob_score = []
KS = []
AUC_A = []
TN = []
FP = []
FN = []
TP = []
for train_index,test_index in kf.split(train_re):
#k += 1
X_train,X_test = train_data[train_index],train_data[test_index] #数据
Y_train,Y_test = train_re[train_index],train_re[test_index]
if balance == 'TRUE':
over_sample = SMOTE(random_state = 255)
over_sample_X,oversample_Y = over_sample.fit_sample(X_train,Y_train)
else:
over_sample_X = X_train
oversample_Y = Y_train
model = RandomForestRegressor(n_estimators=500,oob_score=True)
re_fit = model.fit(over_sample_X,oversample_Y)
re_pred = re_fit.predict(X_test)
re_pred_A = re_pred
re_pred[re_pred>=0.3] = 1
re_pred[re_pred<0.3] = 0
re_pred = re_pred.tolist()
Y_test = Y_test.tolist()
tn1, fp1, fn1, tp1 = confusion_matrix(Y_test, re_pred).ravel()
accuary = (float(tn1)+float(tp1))/(float(tn1)+float(fp1)+float(fn1)+float(tp1))
oobscore = model.oob_score_
K_S = float(tp1)/(float(tp1)+float(fn1))-float(fp1)/(float(fp1)+float(tn1))
AUC = roc_auc_score(Y_test, re_pred_A)
print(accuary)
print(oobscore)
print(K_S)
print(AUC)
predict_score.append(accuary)
oob_score.append(oobscore)
KS.append(K_S)
AUC_A.append(AUC)
TN.append(tn1)
FP.append(fp1)
FN.append(fn1)
TP.append(tp1)
save(TN,FP,FN,TP,predict_score,oob_score,KS,AUC_A,'para_phase_%s.xlsx'%(balance))
print ("success!")