之前用 R 语言一直感觉 .Rdata 格式的文件很好用，可以把每次执行的中间文件保存便于下次调用，刚熟悉 Python 还没接触这块知识，所以有时候做项目不太顺手，索性上网搜了下，整理如下：

模型存档

#############

# joblib 库

from sklearn.linear_model import LogisticRegression

from sklearn.externals import joblib

# 模型保存

lr_model = LogisticRegression()

joblib.dump(lr_model, 'xx.model')

# 模型载入

lr_model = joblib.load('xx.model')

#############

# pickle 库

# 模型保存

import pickle

with open('lr_model.pickle', 'wb') as fp:

pickle.dump(lr_model, fp)

# 模型载入

with open('lr_model.pickle', 'rb') as fp:

lr_model = pickle.load(fp)

命令行参数解析

import argparse

# 定义参数解析

def build_arg_parser():

parser = argparse.ArgumentParser(description='Trains the classifier')

parser.add_argument("--feature-map-file", dest="feature_map_file", required=True,

help="Input pickle file containing the feature map")

parser.add_argument("--model-file", dest="model_file", required=False,

help="Output file where the trained model will be stored")

return parser

args = build_arg_parser().parse_args()

feature_map_file = args.feature_map_file

model_file = args.model_file

对象保存

#############

# pickle 库

import pickle

x, y = 1, range(10)

with open('xx.pickle', 'wb') as fp:

# 通过传递protocol = -1到dump()来减少文件大小

pickle.dump([x, y], fp)

# 对象载入

with open('xx.pickle', 'rb') as fp:

x, y = pickle.load(fp)

print(x)

#############

# _pickle 库

# 对象保存

import _pickle as cpickle

x, y = 1, range(10)

with open('xx.pickle', 'wb') as fp:

# 通过传递protocol = -1到dump()来减少文件大小

cpickle.dump([x, y], fp)

# 对象载入

del x, y

with open('xx.pickle', 'rb') as fp:

x, y = cpickle.load(fp)

print(x)

############

# dill 库

import dill

# 文件保存

filename = 'globalsave.pkl'

dill.dump_session(filename)

# 文件载入

dill.load_session(filename)

############

# 其他库如：pmml，shelve