Pickle模块可以序列化对象并保存到磁盘中，并在需要的时候读取出来，任何对象都可以执行序列化操作。在机器学习中，我们常常需要把训练好的模型存储起来，这样在进行决策时直接将模型独处，而不需要重新训练模型，这样就大大节约了时间。
pickle模块常用函数
| dump(obj,file,[,protocol]) | 将obj对象序列化存入已经打开的file中 |
| load(file) | 将file中的对象序列化读出 |
| dumps(obj,[,protocol]) | 将obj对象序列化为string形式，而不是存入文件中 |
| loads(string) | 从string中读出序列化前的obj对象 |
示例
#coding=utf-8 import pickle datalist = [[1, 1, ‘yes‘], [1, 1, ‘yes‘], [1, 0, ‘no‘], [0, 1, ‘no‘], [0, 1, ‘no‘]] datadict = { 0: [1, 2, 3, 4], 1: (‘a‘, ‘b‘), 2: {‘c‘:‘yes‘,‘d‘:‘no‘}} with open("pickle_test.txt","wb") as writefp: pickle.dump(datalist, writefp) pickle.dump(datadict, writefp) with open("pickle_test.txt", "rb") as readfp: data1 = pickle.load(readfp) data2 = pickle.load(readfp) print (data1) print (data2) p = pickle.dumps(datalist) print( pickle.loads(p) ) p = pickle.dumps(datadict) print( pickle.loads(p) )
>>> [[1, 1, ‘yes‘], [1, 1, ‘yes‘], [1, 0, ‘no‘], [0, 1, ‘no‘], [0, 1, ‘no‘]]
>>> {0: [1, 2, 3, 4], 1: (‘a‘, ‘b‘), 2: {‘c‘: ‘yes‘, ‘d‘: ‘no‘}}
>>> [[1, 1, ‘yes‘], [1, 1, ‘yes‘], [1, 0, ‘no‘], [0, 1, ‘no‘], [0, 1, ‘no‘]]
>>> {0: [1, 2, 3, 4], 1: (‘a‘, ‘b‘), 2: {‘c‘: ‘yes‘, ‘d‘: ‘no‘}}
dump和load相比dumps和loads还有另外一种能力：dump()函数能一个接着一个的将几个对象序列化存储到同一个文件中，随后调用load()来以同样的顺序反序列化读出这些对象