Pytorch builds neural network to predict temperature

import numpy as np
import pandas as pd 
import matplotlib.pyplot as plt
import torch
import torch.optim as optim
import warnings
warnings.filterwarnings("ignore")
%matplotlib inline


features = pd.read_csv('temps.csv')

# See what the data looks like 
features.head()

#  Processing time data ： Convert all time columns into datetime Format , Convenient visual display 
import datetime

#  Get years respectively , month , Japan 
years = features['year']
months = features['month']
days = features['day']

# datetime Format 
dates = [str(int(year)) + '-' + str(int(month)) + '-' + str(int(day)) for year, month, day in zip(years, months, days)]
dates = [datetime.datetime.strptime(date, '%Y-%m-%d') for date in dates]

#  Prepare to draw 
#  Specify the default style 
plt.style.use('fivethirtyeight')

#  Setting up layout 
fig, ((ax1, ax2), (ax3, ax4)) = plt.subplots(nrows=2, ncols=2, figsize = (10,10))
fig.autofmt_xdate(rotation = 45)

#  Label value 
ax1.plot(dates, features['actual'])
ax1.set_xlabel(''); ax1.set_ylabel('Temperature'); ax1.set_title('Max Temp')

#  yesterday 
ax2.plot(dates, features['temp_1'])
ax2.set_xlabel(''); ax2.set_ylabel('Temperature'); ax2.set_title('Previous Max Temp')

#  The day before yesterday 
ax3.plot(dates, features['temp_2'])
ax3.set_xlabel('Date'); ax3.set_ylabel('Temperature'); ax3.set_title('Two Days Prior Max Temp')

#  My funny friend 
ax4.plot(dates, features['friend'])
ax4.set_xlabel('Date'); ax4.set_ylabel('Temperature'); ax4.set_title('Friend Estimate')

plt.tight_layout(pad=2)

#  Hot coding alone   Right week 1 Code on Sunday   Convert all data into numerical form 
features = pd.get_dummies(features)# Send the whole data in pd.get_dummies It will automatically judge which column is a string and can be encoded 
features.head(5)

#  Specify the label to be trained before the training model X Y
labels = np.array(features['actual'])# take y value 

#  Remove labels from features 
features= features.drop('actual', axis = 1)# take x value 

#  Save the names separately , For visual display 
feature_list = list(features.columns)

#  Convert to appropriate format 
features = np.array(features)

from sklearn import preprocessing
input_features = preprocessing.StandardScaler().fit_transform(features)# The data value difference between some columns is relatively large, and data standardization is needed for preprocessing , The convergence speed is faster after standardization , The loss value is smaller 

torch Build a network model

# First of all, x y Turn into tensor The format of 
x = torch.tensor(input_features, dtype = float)

y = torch.tensor(labels, dtype = float)

#  Weight parameter initialization 
weights = torch.randn((14, 128), dtype = float, requires_grad = True) # Cryptic layer 128 Neurons , The input layer passes through the hidden layer to get 128 Features 
biases = torch.randn(128, dtype = float, requires_grad = True) 
weights2 = torch.randn((128, 1), dtype = float, requires_grad = True) # Because it's a return mission , So finally, we need to get a value to represent w, So we need to go through a full connection layer [128,1]
biases2 = torch.randn(1, dtype = float, requires_grad = True) 

learning_rate = 0.001 
losses = []

for i in range(1000):
    #  Calculate hidden layer 
    hidden = x.mm(weights) + biases#[348,14] *[14,128] x.mm() Representation matrix multiplication 
    #  Add activation function 
    hidden = torch.relu(hidden)
    #  Predicted results 
    predictions = hidden.mm(weights2) + biases2
    #  Calculate the loss by 
    loss = torch.mean((predictions - y) ** 2) 
    losses.append(loss.data.numpy())
    
    #  Print loss value 
    if i % 100 == 0:
        print('loss:', loss)
    # Backward propagation calculation 
    loss.backward()
    
    # Update parameters 
    weights.data.add_(- learning_rate * weights.grad.data)  # The minus sign means the gradient drops 
    biases.data.add_(- learning_rate * biases.grad.data)
    weights2.data.add_(- learning_rate * weights2.grad.data)
    biases2.data.add_(- learning_rate * biases2.grad.data)
    
    #  Remember to empty every iteration 
    weights.grad.data.zero_()
    biases.grad.data.zero_()
    weights2.grad.data.zero_()
    biases2.grad.data.zero_()

It's simpler to build a network model

The first is to calculate all the input data together , Next use miniBatch Small batch calculation

input_size = input_features.shape[1]
hidden_size = 128
output_size = 1
batch_size = 16
# Here, the network architecture is built in advance 
my_nn = torch.nn.Sequential(
    torch.nn.Linear(input_size, hidden_size),
    torch.nn.Sigmoid(),
    torch.nn.Linear(hidden_size, output_size),
)
cost = torch.nn.MSELoss(reduction='mean')
optimizer = torch.optim.Adam(my_nn.parameters(), lr = 0.001)# The simple method replaces the previous need for handwriting gradient descent and gradient update   The loss function is also called directly MSE Just lose 
# The optimizer uses ADam The learning rate can be adjusted dynamically , Adding attenuation function makes the learning rate slowly decrease as the iteration proceeds 

#  Training network 
losses = []
for i in range(1000):
    batch_loss = []
    # MINI-Batch Methods to train 
    for start in range(0, len(input_features), batch_size):
        end = start + batch_size if start + batch_size < len(input_features) else len(input_features)
        xx = torch.tensor(input_features[start:end], dtype = torch.float, requires_grad = True)
        yy = torch.tensor(labels[start:end], dtype = torch.float, requires_grad = True)
        prediction = my_nn(xx)# Forward propagation prediction 
        loss = cost(prediction, yy)# Calculate the loss 
        optimizer.zero_grad()# Gradient clear 
        loss.backward(retain_graph=True)# Back propagation 
        optimizer.step()# Gradient update 
        batch_loss.append(loss.data.numpy())
    
    #  Print loss 
    if i % 100==0:
        losses.append(np.mean(batch_loss))
        print(i, np.mean(batch_loss))

Predict training results

x = torch.tensor(input_features, dtype = torch.float)
predict = my_nn(x).data.numpy()# Turn into numpy1 Grid is convenient for drawing 

#  Convert date format 
dates = [str(int(year)) + '-' + str(int(month)) + '-' + str(int(day)) for year, month, day in zip(years, months, days)]
dates = [datetime.datetime.strptime(date, '%Y-%m-%d') for date in dates]

#  Create a table to store dates and their corresponding tag values 
true_data = pd.DataFrame(data = {
    'date': dates, 'actual': labels})

#  Empathy , Create another one to save the date and its corresponding model prediction value 
months = features[:, feature_list.index('month')]
days = features[:, feature_list.index('day')]
years = features[:, feature_list.index('year')]

test_dates = [str(int(year)) + '-' + str(int(month)) + '-' + str(int(day)) for year, month, day in zip(years, months, days)]

test_dates = [datetime.datetime.strptime(date, '%Y-%m-%d') for date in test_dates]

predictions_data = pd.DataFrame(data = {
    'date': test_dates, 'prediction': predict.reshape(-1)}) #predict.reshape(-1) Means a column , Cannot be in matrix format 

#  True value 
plt.plot(true_data['date'], true_data['actual'], 'b-', label = 'actual')

#  Predictive value 
plt.plot(predictions_data['date'], predictions_data['prediction'], 'ro', label = 'prediction')
plt.xticks(rotation = '60'); 
plt.legend()

#  Title 
plt.xlabel('Date'); plt.ylabel('Maximum Temperature (F)'); plt.title('Actual and Predicted Values');

The experimental data set is transformed into .csv Files use