数据分析代码

import pandas as pd
%matplotlib notebook

data = pd.read_csv('',header=None)
data.columns=['ip','interface','time','In_Bytes','E_Bytes','In_Packets','E_Packets']

int_data = data[(data['ip'] == '129.60.161.169')&(data['interface'] == 'xgei-0/2/0/24')]
int_data.sort_values('time',inplace=True)
int_data.reset_index(inplace=True,drop=True)
int_data['diff_in'] = int_data['In_Bytes'].diff()/1024
int_data['diff_out'] = int_data['E_Bytes'].diff()/1024
int_data.fillna(0,inplace=True)
int_data[:200].plot('time',['diff_in'],kind='scatter')

# 非0数据比例
sum(int_data['diff_in']>0)/len(int_data)


int_data.describe()


# 数据归一化
int_data['E_Bytes'] = (int_data['E_Bytes']-int_data['E_Bytes'].min())/(int_data['E_Bytes'].max()--int_data['E_Bytes'].min())
int_data['In_Bytes'] = (int_data['In_Bytes']-int_data['In_Bytes'].min())/(int_data['In_Bytes'].max()--int_data['In_Bytes'].min())
int_data.plot('time',['E_Bytes','In_Bytes'],kind='line')


# 协方差：绝对值越大，线性关系越强
print('协方差：',int_data['diff_in'].cov(int_data['diff_out']))
# 相关系数：相关系数在-1到1之间，接近1为正相关，接近-1为负相关，0为不相关。
print('相关系数：',int_data['diff_in'].corr(int_data['diff_out']))


int_data.plot('time',['diff_in','diff_out'],kind='line')


int_data['diff_in'] = (int_data['diff_in']-int_data['diff_in'].min())/(int_data['diff_in'].max()--int_data['diff_in'].min())
int_data['diff_out'] = (int_data['diff_out']-int_data['diff_out'].min())/(int_data['diff_out'].max()--int_data['diff_out'].min())
int_data.plot('time',['diff_in','diff_out'],kind='line')

import seaborn as sns
import matplotlib.pyplot as plot
%matplotlib inline
sns.set(context='notebook', style='darkgrid', palette='deep', font='sans-serif', font_scale=1, color_codes=True, rc=None)
for interface in list(data[data['ip'] == '129.60.161.169']['interface'].value_counts().index):
    t = data[(data['ip'] == '129.60.161.169') & (data['interface'] == interface)]['In_Bytes']
    print(t.values)
    ax = sns.lineplot(data=t.values)# 默认绘制折线图