<?xml version="1.0" encoding="utf-8" ?><rss version="2.0"><channel><title><![CDATA[jack_ricky的博客]]></title><description><![CDATA[]]></description><link>https://blog.youkuaiyun.com/jack_ricky</link><language>zh-cn</language><generator>https://blog.youkuaiyun.com/</generator><copyright><![CDATA[Copyright &copy; jack_ricky]]></copyright><item><title><![CDATA[Python 常用（聚类/分类）数据预处理方法]]></title><link>https://blog.youkuaiyun.com/jack_ricky/article/details/100040153</link><guid>https://blog.youkuaiyun.com/jack_ricky/article/details/100040153</guid><author>jack_ricky</author><pubDate>Thu, 12 Dec 2019 10:09:00 +0800</pubDate><description><![CDATA[要对数据进行聚类/分类，往往需要经过以下几个步骤：

处理缺失值（我比较少遇到）--》数据标准化 --》降维（白化） --》训练 --》预测 --》评价模型效果

0. 缺失值的处理一般是填充缺省值或者平均值等方法，我一般直接补0

1. 对数据进行标准化。常用的方法有z-score和maxmin。其中，z-socre是(x-mean)/std,也就是去中心化。注意：标准化方法（以及降维）只能在训...]]></description><category></category></item><item><title><![CDATA[使用TeamCymru获得IP对应的BGP-Prefix和AS number]]></title><link>https://blog.youkuaiyun.com/jack_ricky/article/details/103075303</link><guid>https://blog.youkuaiyun.com/jack_ricky/article/details/103075303</guid><author>jack_ricky</author><pubDate>Thu, 14 Nov 2019 21:35:38 +0800</pubDate><description><![CDATA[将所有的IP分成2500个IP一个文件(split -l 2500 all_ip.txt)，每个文件第一二行分别是begin,verbose
	使用TeamCymru提供的接口处理每个文件
	
#!/bin/bash

PROCESS_NUM=0
function getdir(){
    for element in `ls $1`
    do  
        dir_or_file=$...]]></description><category></category></item><item><title><![CDATA[Python 读取文件的几种方法]]></title><link>https://blog.youkuaiyun.com/jack_ricky/article/details/102958734</link><guid>https://blog.youkuaiyun.com/jack_ricky/article/details/102958734</guid><author>jack_ricky</author><pubDate>Thu, 07 Nov 2019 17:10:24 +0800</pubDate><description><![CDATA[0. 读取文件


# 1
f = open('fileName', 'r')
# do something with f
f.close()

# 2
with open('fileName', 'r') as f:
    # do something with f
    # no need to close, and will close automatically if exceptio...]]></description><category></category></item><item><title><![CDATA[Python 内存回收]]></title><link>https://blog.youkuaiyun.com/jack_ricky/article/details/99967968</link><guid>https://blog.youkuaiyun.com/jack_ricky/article/details/99967968</guid><author>jack_ricky</author><pubDate>Wed, 21 Aug 2019 15:40:35 +0800</pubDate><description><![CDATA[import gc

del data
gc.collect()


]]></description><category></category></item><item><title><![CDATA[Linux没有root权限时，pip安装Python模块]]></title><link>https://blog.youkuaiyun.com/jack_ricky/article/details/96973377</link><guid>https://blog.youkuaiyun.com/jack_ricky/article/details/96973377</guid><author>jack_ricky</author><pubDate>Tue, 23 Jul 2019 10:40:36 +0800</pubDate><description><![CDATA[首先在https://pypi.org/上获得所需模块的.whl文件。然后：

pip install ***.whl
]]></description><category></category></item><item><title><![CDATA[MATLAB GMM]]></title><link>https://blog.youkuaiyun.com/jack_ricky/article/details/95178635</link><guid>https://blog.youkuaiyun.com/jack_ricky/article/details/95178635</guid><author>jack_ricky</author><pubDate>Tue, 09 Jul 2019 11:25:52 +0800</pubDate><description><![CDATA[clear
clc

disp 'Loading...'
feature_list = csvread('feature_list.csv');
disp 'scaling...'
feature_z = zscore(feature_list);


GMM_3 = fitgmdist(feature_z, 3, 'RegularizationValue', 0.1);
disp 'GMM 3...]]></description><category></category></item><item><title><![CDATA[linux环境下 python import找不到自定义的模块]]></title><link>https://blog.youkuaiyun.com/jack_ricky/article/details/93461717</link><guid>https://blog.youkuaiyun.com/jack_ricky/article/details/93461717</guid><author>jack_ricky</author><pubDate>Mon, 24 Jun 2019 09:32:35 +0800</pubDate><description><![CDATA[两种方法：

1. 在导入模块之前（每个.py文件都需要加）


import sys
sys.path.append(r'/home/my_python_module/')

2. 更改环境变量（一次修改永久生效）

管理员权限修改：/etc/profile。在最后一行加上：


export PYTHONPATH=$PYTHONPATH:/home/my_python_module/

然后运...]]></description><category></category></item><item><title><![CDATA[Holt-Winters 方法进行时间序列预测 与 评价指标]]></title><link>https://blog.youkuaiyun.com/jack_ricky/article/details/88982170</link><guid>https://blog.youkuaiyun.com/jack_ricky/article/details/88982170</guid><author>jack_ricky</author><pubDate>Tue, 02 Apr 2019 21:51:29 +0800</pubDate><description><![CDATA[本文主要对HW方法进行简单的特点、使用介绍。

HW方法比较适用于具有明确周期性的序列。python代码如下：


from statsmodels.tsa.holtwinters import ExponentialSmoothing 

def hw_predict(sample_serie, s_p=4, s_index=0, e_index=10):
    hw_fit = Expone...]]></description><category></category></item><item><title><![CDATA[【MATLAB】 画 GMM component 曲线 & 计算曲线的交点]]></title><link>https://blog.youkuaiyun.com/jack_ricky/article/details/88643061</link><guid>https://blog.youkuaiyun.com/jack_ricky/article/details/88643061</guid><author>jack_ricky</author><pubDate>Mon, 18 Mar 2019 17:02:03 +0800</pubDate><description><![CDATA[xgrid = linspace(0,1,1000)';
n1 = makedist('normal',GMM.mu(1),sqrt(GMM.Sigma(1))); % make distribution
n2 = makedist('normal',GMM.mu(2),sqrt(GMM.Sigma(2)));
p = GMM.ComponentProportion;
y_1 = p(1)*pd...]]></description><category></category></item><item><title><![CDATA[Linux 解压.gz文件]]></title><link>https://blog.youkuaiyun.com/jack_ricky/article/details/88239624</link><guid>https://blog.youkuaiyun.com/jack_ricky/article/details/88239624</guid><author>jack_ricky</author><pubDate>Wed, 06 Mar 2019 15:57:38 +0800</pubDate><description><![CDATA[//分别将exp1.txt和exp2.txt压缩，且不保留原文件
gzip exp1.txt exp2.txt　　　　 


//将exp1.gz解压的结果放置在文件exp.1中，并且原压缩文件exp1.gz不会消失
gzip -cd exp1.gz &amp;gt; exp.1


...]]></description><category></category></item><item><title><![CDATA[Python 正则表达式使用]]></title><link>https://blog.youkuaiyun.com/jack_ricky/article/details/88222789</link><guid>https://blog.youkuaiyun.com/jack_ricky/article/details/88222789</guid><author>jack_ricky</author><pubDate>Wed, 06 Mar 2019 10:56:01 +0800</pubDate><description><![CDATA[import re

regex = re.compile(r'(uid|sid)=(.+?)&amp;amp;') # what in '' is the reg exp we want.

f = open(&quot;log.txt&quot;, &quot;r&quot;)
logs = f.readlines()

for log in logs:
    kvs = regex.findall(log)
    for kv in...]]></description><category></category></item><item><title><![CDATA[Python/Bash 遍历文件夹]]></title><link>https://blog.youkuaiyun.com/jack_ricky/article/details/88222429</link><guid>https://blog.youkuaiyun.com/jack_ricky/article/details/88222429</guid><author>jack_ricky</author><pubDate>Wed, 06 Mar 2019 10:42:59 +0800</pubDate><description><![CDATA[import os

infiles = list() # the absolute path we want
indir   = './root_data/'

for root, dirs, files in os.walk(indir):
    for filename in files:
        infiles.append(os.path.join(root, filenam...]]></description><category></category></item><item><title><![CDATA[Python matplotlib 以pdf形式保存图片]]></title><link>https://blog.youkuaiyun.com/jack_ricky/article/details/88073324</link><guid>https://blog.youkuaiyun.com/jack_ricky/article/details/88073324</guid><author>jack_ricky</author><pubDate>Sat, 02 Mar 2019 11:04:21 +0800</pubDate><description><![CDATA[import matplotlib.pyplot as plt 
from matplotlib.backends.backend_pdf import PdfPages

pdf = PdfPages('des_file_name.pdf')

plt.figure()
plt.hist(data)
plt.tight_layout()

print 'savefig...'
pdf.save...]]></description><category></category></item><item><title><![CDATA[Python 多进程]]></title><link>https://blog.youkuaiyun.com/jack_ricky/article/details/88073258</link><guid>https://blog.youkuaiyun.com/jack_ricky/article/details/88073258</guid><author>jack_ricky</author><pubDate>Sat, 02 Mar 2019 10:59:49 +0800</pubDate><description><![CDATA[Python的多线程是假的，但是多进程是真的。在处理计算需求比较大的问题时可以考虑使用多进程：


#coding=utf-8
import copy
import numpy as np
import multiprocessing 
import z_facility as fac
from multiprocessing import Pool

# 输入处理的一个单元，比如list的一个元...]]></description><category></category></item><item><title><![CDATA[python 带条纹的 barplot 和 boxplot]]></title><link>https://blog.youkuaiyun.com/jack_ricky/article/details/87733990</link><guid>https://blog.youkuaiyun.com/jack_ricky/article/details/87733990</guid><author>jack_ricky</author><pubDate>Tue, 19 Feb 2019 21:03:28 +0800</pubDate><description><![CDATA[
plt.boxplot(..., patch_artist=True, boxprops=dict(facecolor=&quot;green&quot;, hatch = '/'))

plt.bar(..., hatch='\\')

 ]]></description><category></category></item><item><title><![CDATA[LaTex插入eps或pdf（或svg）以及去除eps、pdf白边]]></title><link>https://blog.youkuaiyun.com/jack_ricky/article/details/83549192</link><guid>https://blog.youkuaiyun.com/jack_ricky/article/details/83549192</guid><author>jack_ricky</author><pubDate>Tue, 30 Oct 2018 21:29:46 +0800</pubDate><description><![CDATA[1. 插入eps


\begin{figure}
\centering
\includegraphics[width=3.33in, keepaspectratio]{fig_name}
\caption{title}
\end{figure}

2. 插入pdf（类似于插入eps）


\begin{figure}
\centering
\includegraphics[width=3.33i...]]></description><category></category></item><item><title><![CDATA[IP转经纬度，经纬度画热力图]]></title><link>https://blog.youkuaiyun.com/jack_ricky/article/details/83415354</link><guid>https://blog.youkuaiyun.com/jack_ricky/article/details/83415354</guid><author>jack_ricky</author><pubDate>Fri, 26 Oct 2018 15:13:59 +0800</pubDate><description><![CDATA[1. IP转经纬度

工具：GeoIP python
	每个IP一行
	代码：

import GeoIP
import sys
import os

def print_usage():
 print &quot;Usage: &quot; + os.path.basename(__file__) + &quot; ipfile.txt&quot;

# make sure we've been passed the filename...]]></description><category></category></item><item><title><![CDATA[sed/awk删除文件奇数行或偶数行]]></title><link>https://blog.youkuaiyun.com/jack_ricky/article/details/82179623</link><guid>https://blog.youkuaiyun.com/jack_ricky/article/details/82179623</guid><author>jack_ricky</author><pubDate>Wed, 29 Aug 2018 10:45:49 +0800</pubDate><description><![CDATA[ 


sed '1~2d'  file &amp;gt; new_file # 删除奇数行
sed '1~2!d' file &amp;gt; new_file # 删除偶数行

awk 'NR%2==0' filename &amp;gt; new_file # 删除奇数行
awk 'NR%2==1' filename &amp;gt; new_file # 删除偶数行

 ...]]></description><category></category></item><item><title><![CDATA[python画双y轴图像]]></title><link>https://blog.youkuaiyun.com/jack_ricky/article/details/81939208</link><guid>https://blog.youkuaiyun.com/jack_ricky/article/details/81939208</guid><author>jack_ricky</author><pubDate>Wed, 22 Aug 2018 10:26:09 +0800</pubDate><description><![CDATA[很多时候可能需要在一个图中画出多条函数图像，但是可能y轴的物理含义不一样，或是数值范围相差较大，此时就需要双y轴。

matplotlib和seaborn都可以画双y轴图像。一个例子：


import seaborn as sns
import matplotlib.pyplot as plt

# ax1 for KDE, ax2 for CDF

f, ax1 = plt.subplots(...]]></description><category></category></item><item><title><![CDATA[Python画loglog CCDF]]></title><link>https://blog.youkuaiyun.com/jack_ricky/article/details/81838583</link><guid>https://blog.youkuaiyun.com/jack_ricky/article/details/81838583</guid><author>jack_ricky</author><pubDate>Sun, 19 Aug 2018 15:50:00 +0800</pubDate><description><![CDATA[
data.sort()
p = 1. * arange(len(data)) / (len(data)-1) # 计算各点的累计概率 F(x)
p = [1-i for i in p]                       # 计算概率的补 1-F(x)
x = np.log10(data)                         # logx
y = np.log10(p)   ...]]></description><category></category></item></channel></rss>