Python应用程序调优

最近在做一些深度学习、关联规则挖掘相关的算法，其中运行时间和占用内存是算法好坏的评价标准之二，在算法实现过程中踩了一些坑，也误打误撞找到了一些解决办法，将其发布于本博客，本博客将持续更新!!!

1 时间调优

1.1 不要在程序内部频繁使用垃圾回收函数 gc.collect()，这将使程序的运行时间慢100倍

# 示例程序
start = time.time()
lines = ['X1','X2','X3','X4','X5','X6','X7']
line = ['X3','X5','X7']
for ln in lines:
    if ln in line:
        ln = ln
del lines, line
gc.collect()
end = time.time()
print("使用垃圾回收: %sS"%(end-start))

1.2 判断一个列表是否在另一个列表中时直接用循环代替issubset()

# 示例程序
start = time.time()
lines = ['X1','X2','X3','X4','X5','X6','X7']
line = ['X3','X5','X7']
flag = True
for ln in line:
    if ln not in lines:
        flag = False
        break
if flag:
    print('Find')
end = time.time()
print("循环判断时: %sS"%(end-start))

start = time.time()
lines = ['X1','X2','X3','X4','X5','X6','X7']
line = ['X3','X5','X7']
if set(line).issubset(set(lines)):
    print('Find')
end = time.time()
print("使用issubset()判断时: %sS"%(end-start))

1.3 采用itertools函数生成全排列时使用combinations()不要使用product()，后者比前者慢100倍

# 示例程序
line = ['X10', 'X9', 'X2', 'X6', 'X4', 'X5']
count = len(line)

lines = []
s = time.time()
for i in range(2,count):
    lines.append(list(combinations(line, i)))
e = time.time()
print('Combinations:{:f}'.format(e-s))
s = time.time()
for i in range(2,count):
    lines.append(list(product(line, repeat=i)))
e = time.time()
print('Product:{:f}'.format(e-s))

未完待续......

2. 内存调优

未完待续......