pytorch搭建卷积神经网络【第十课_tensor_# 时间序列数据操作_4】

---

前言

案例代码https://github.com/2012Netsky/pytorch_cnn/blob/main/4_time_series_bikes.ipynb

一、tensor打印配置

#!/usr/bin/env python
# coding: utf-8
import numpy as np
import torch
torch.set_printoptions(edgeitems=2, threshold=50)
# precision是每一个元素的输出精度，默认是八位；
# threshold是输出时的阈值，当tensor中元素的个数大于该值时，进行缩略输出，默认时1000；
# edgeitems是输出的维度，默认是3；
# linewidth字面意思，每一行输出的长度；
# profile=None，修正默认设置（不太懂，感兴趣的可以试试）

二、读取txt

bikes_numpy = np.loadtxt(
    "../data/p1ch4/bike-sharing-dataset/hour-fixed.csv", 
    dtype=np.float32, 
    delimiter=",", 
    skiprows=1, 
    converters={1: lambda x: float(x[8:10])}) # <1>只显示日期 年、月去掉
bikes = torch.from_numpy(bikes_numpy)
print(bikes[:,1])

bikes.shape, bikes.stride()

三、 重构维度

# （-1，24，17） -1为自动计算剩余维数
daily_bikes = bikes.view(-1, 24, bikes.shape[1]) 
daily_bikes.shape, daily_bikes.stride()# (408=17*24, 17=1*17, 1）

# 交换维度 0 1 2 三个维度交换
daily_bikes = daily_bikes.transpose(1, 2)# 1维度和2维度交换
daily_bikes.shape, daily_bikes.stride()

四、提取相关数据

first_day = bikes[:24].long()# 前24个
weather_onehot = torch.zeros(first_day.shape[0], 4)
print(bikes.shape)
print(first_day.shape)
print(first_day[:,9])# 第九列

print(weather_onehot.shape)

print(weather_onehot.shape)
print(weather_onehot)

五、独热编码

# 独热编码
weather_onehot1 = weather_onehot.scatter_(
    dim=1, 
    index=first_day[:,9].unsqueeze(1).long() - 1, # <1>
    value=1.0)
torch.set_printoptions(profile="full")

print(weather_onehot1.shape)
print(weather_onehot1)

六、组合与拆分tensor

#torch.cat是将两个张量（tensor）拼接在一起，cat是concatnate的意思，即拼接，联系在一起。
# C = torch.cat( (A,B),0 )  #按维数0拼接（竖着拼）
# C = torch.cat( (A,B),1 )  #按维数1拼接（横着拼）
torch.cat((bikes[:24], weather_onehot), 1)[:1]

print(bikes.shape)
print()

print(bikes[:24].shape)
print(weather_onehot.shape)
print((torch.cat((bikes[:24], weather_onehot), 1)).shape)
print()

# 所有行 第一列
print((torch.cat((bikes[:24], weather_onehot), 1)[:1]).shape) 

daily_weather_onehot = torch.zeros(daily_bikes.shape[0], 4,
                                   daily_bikes.shape[2])
daily_weather_onehot.shape

print(daily_bikes.shape[0], daily_bikes.shape[2])
print(daily_bikes[:,9,:].shape)

# 独热编码
daily_weather_onehot.scatter_(
    1, daily_bikes[:,9,:].long().unsqueeze(1) - 1, 1.0)
daily_weather_onehot.shape

daily_bikes = torch.cat((daily_bikes, daily_weather_onehot), dim=1)

print(daily_bikes.shape)
print(daily_weather_onehot.shape)
print(( torch.cat((daily_bikes, daily_weather_onehot), dim=1)).shape)

七、正则化数据

daily_bikes[:, 9, :] = (daily_bikes[:, 9, :] - 1.0) / 3.0

# 变量标准化 对训练有益
# 变量标准化 方法1 映射到范围0，1
temp = daily_bikes[:, 10, :]
temp_min = torch.min(temp)
temp_max = torch.max(temp)
daily_bikes[:, 10, :] = ((daily_bikes[:, 10, :] - temp_min)
                         / (temp_max - temp_min))


# 变量标准化 方法1 减去均值除以标准差
temp = daily_bikes[:, 10, :]
daily_bikes[:, 10, :] = ((daily_bikes[:, 10, :] - torch.mean(temp))
                         / torch.std(temp))

总结