对较大的原始csv文件抽取出一部分样本

避免对大文件全部读取到内存中，浪费时间，也能避免内存溢出；
先对文件先进行抽样，抽出很小一部分，测试程序的语法正确性，再用全部文件测试程序的功能正确性；

import tensorflow as tf
import numpy as np
import pandas as pd
import matplotlib.pyplot as plt
import csv

# 原始csv文件包含222个类，每个类500个样本，连续排列
num_sample = 10;
with open("data/clean_data/normalized_training.csv", "rb") as fi:
    with open("data/clean_data/normalized_training_part.csv",'wb') as fo:
        fo.write(fi.readline()) # 读取csv文件column name行，也就是第一行
        for i in range(222):
            for j in range(num_sample):
                fo.write(fi.readline())
            for j in range(500-num_sample):
                fi.readline();

 

num_sample = 10;with open("data/clean_data/normalized_training.csv", "rb") as fi:    with open("data/clean_data/normalized_training_part.csv",'wb') as fo:        fo.write(fi.readline())        for i in range(222):            for j in range(num_sample):                fo.write(fi.readline())            for j in range(500-num_sample):                fi.readline();