Python中os.walk()的使用方法 - 晓伟的文章 - 知乎
https://zhuanlan.zhihu.com/p/149824829
如何重置index？
data.reset_index(drop=True,inplace = True) ## 先把之前的index去掉~
time_list = pd.date_range(start = "2019/12/10 17:30",end = "2019/12/24 21:59:59.8",freq = "0.2S")
data.index.name = "time_"
data = data.set_index(time_list)
如果index是日期加时间，列是通量，那么如何获得每个小时的平均值？
data["hour"] = pd.to_datetime(data.index).hour
transform = lambda x:x.to_pydatetime().replace(hour=0)
data = data.set_index([data.index,data['hour']])
cal_q=lambda x:x.describe(percentiles=[0.05,0.25,.5,.75, .95])
data.loc[:,"Flux"].unstack().apply(cal_q).to_csv("Flux.csv")
如果想对长数据进行切片并输出？
import numpy as np
import pandas as pd
import matplotlib as mpl
import os
import math
import glob
import datetime
import matplotlib.dates as mdates
from matplotlib import pyplot as plt
from datetime import timedelta
def control_data_level(data,n):
time_list = pd.date_range(start='2019-12-24 22:00',end='2020-1-1 6:30',freq='30T')
for ith, star_t in enumerate(time_list[:]):
end_t = star_t + timedelta(minutes=30)-timedelta(seconds=0.2)
print(star_t, end_t)
sub_df = data.loc[star_t:end_t]
sub_df.to_csv(r"D:\python\inter-1-"+datetime.datetime.strftime(star_t,'%Y-%m-%d_%H%M')+'.csv')
fig,axs = plt.subplots(nrows=6,ncols=1,figsize=(30,20))
mpl.rcParams['font.size'] = 15
mpl.rcParams['font.weight'] = 'bold'
mpl.rcParams['font.sans-serif']=['Arial']
axs[0].plot(sub_df.index,sub_df["SCsize_tmp"],"k-o",lw=1,markerfacecolor='w')
axs[1].plot(sub_df.index,sub_df["SizeBC_tmp"],"b-o",lw=1,markerfacecolor='w')
axs[2].plot(sub_df.index,sub_df["massBC"],"r-o",lw=1,markerfacecolor='w')
axs[3].plot(sub_df.index,sub_df["Ux_12m"],"k-o",lw=1,markerfacecolor='w')
axs[4].plot(sub_df.index,sub_df["Uy_12m"],"b-o",lw=1,markerfacecolor='w')
axs[5].plot(sub_df.index,sub_df["Uz_12m"],"r-o",lw=1,markerfacecolor='w')
axs[0].axhline(sub_df["SCsize_tmp"].mean()+3.5*sub_df["SCsize_tmp"].std(),c="r",ls="--",lw=1.5)
axs[0].axhline(sub_df["SCsize_tmp"].mean()-3.5*sub_df["SCsize_tmp"].std(),c="r",ls="--",lw=1.5)
axs[1].axhline(sub_df["SizeBC_tmp"].mean()+3.5*sub_df["SizeBC_tmp"].std(),c="r",ls="--",lw=1.5)
axs[1].axhline(sub_df["SizeBC_tmp"].mean()-3.5*sub_df["SizeBC_tmp"].std(),c="r",ls="--",lw=1.5)
axs[2].axhline(sub_df["massBC"].mean()+3.5*sub_df["massBC"].std(),c="r",ls="--",lw=1.5)
axs[2].axhline(sub_df["massBC"].mean()-3.5*sub_df["massBC"].std(),c="r",ls="--",lw=1.5)
axs[3].axhline(sub_df["Ux_12m"].mean()+3.5*sub_df["Ux_12m"].std(),c="r",ls="--",lw=1.5)
axs[3].axhline(sub_df["Ux_12m"].mean()-3.5*sub_df["Ux_12m"].std(),c="r",ls="--",lw=1.5)
axs[4].axhline(sub_df["Uy_12m"].mean()+3.5*sub_df["Uy_12m"].std(),c="r",ls="--",lw=1.5)
axs[4].axhline(sub_df["Uy_12m"].mean()-3.5*sub_df["Uy_12m"].std(),c="r",ls="--",lw=1.5)
axs[5].axhline(sub_df["Uz_12m"].mean()+5*sub_df["Uz_12m"].std(),c="r",ls="--",lw=1.5)
axs[5].axhline(sub_df["Uz_12m"].mean()-5*sub_df["Uz_12m"].std(),c="r",ls="--",lw=1.5)
axs[5].set_xlabel("time")
axs[0].set_ylabel("SC num conc.(#/(cm$^3$))")
axs[1].set_ylabel("BC num conc.(#/(cm$^3$))")
axs[2].set_ylabel("BC mass conc.(ng/m$^3$)")
axs[3].set_ylabel("u(m/s)")
axs[4].set_ylabel("v(m/s)")
axs[5].set_ylabel("w(m/s)")
for ax in axs:
ax.xaxis.set_major_formatter(mdates.DateFormatter("%Y-%m-%d %H:%M"))
ax.set_xlim(sub_df.index[0],sub_df.index[-1])
outf=r'D:\python\18000-02s-pic-'+datetime.datetime.strftime(star_t,'%Y-%m-%d_%H%M')+'.png'
fig.savefig(outf,bbox_inches = 'tight')
nan_count = data.resample("0.5H").count()
first_mean = data.resample("0.5H").mean()
first_std = data.resample("0.5H").std()
first_median = data.resample("0.5H").median()
writer = pd.ExcelWriter(r'D:\python\18000_02s-describe_'+str(n)+'.xlsx')
nan_count.to_excel(writer, sheet_name="nan_count")
first_mean.to_excel(writer, sheet_name="first_mean")
first_std.to_excel(writer, sheet_name="first_std")
first_median.to_excel(writer, sheet_name="first_median")
writer.save()