1. Intro

  • Data Preparing and Loading

  • ML中的数据可以是任何能想到的事物，如excel spreadsheet，
    任何类型图像，(youtube)视频，音频，蛋白质(protein)结构，文本等；

MLGamePart

  • ML由两部分组成：A：数据；B：选择或构建的模型；

  • 使用线性回归linear regression创建具有已知参数parameter(模型可学习的事物)
    的数据，看是否可使用梯度下降gradient descent来建立模型来评估这些参数；

  • Linear Regression

  • Gradient Descent

  • 对相关术语，无需特别关心，因没太大意义，会看到实际应用；

# create *known* parameter
bias = 0.3
weight = 0.7

# create data
start = 0
end = 1
step = 0.02
X = torch.arange(start,end,step).unsqueeze(dim=1)
y = weight * X + bias

X[:10],y[:10]

  • X：horizontal_feature，y：vertical_label；

(tensor([[0.0000],
         [0.0200],
         [0.0400],
         [0.0600],
         [0.0800],
         [0.1000],
         [0.1200],
         [0.1400],
         [0.1600],
         [0.1800]]),
 tensor([[0.3000],
         [0.3140],
         [0.3280],
         [0.3420],
         [0.3560],
         [0.3700],
         [0.3840],
         [0.3980],
         [0.4120],
         [0.4260]]))

2. Split Data

  • split data into training and test set

  • 注：处理真实数据时，此步骤通常在项目开始时完成，测试集应始终与所有其它数据分开；
    希望模型能学习训练数据，然后在测试数据上对其评估，
    以了解它对未见过示例example的泛化generalize程度

Type Percentage Purpose

TrainingSet

60% ~ 80%

模型从数据中学习，如在学期semester中学习的课程资料

ValidationSet

10% ~ 20%

模型根据数据进行调整，如模拟考试

TestingSet

10% ~ 20%

模型根据数据进行评估，以测试学到的知识，如期末考试

 
import torch

# nn contain all pytorch building block for neural network
from torch import nn
import matplotlib.pyplot as plt

# check PyTorch version
torchVersion = torch.__version__

# create *known* parameter
bias = 0.3
weight = 0.7

# create data
start = 0
end = 1
step = 0.02
X = torch.arange(start,end,step).unsqueeze(dim=1)
y = weight * X + bias

X[:10],y[:10]

# create training(80%)/testing(20%) split
train_split = int(0.8 * len(X))
X_train,y_train = X[:train_split],y[:train_split]
X_test,y_test = X[train_split:],y[train_split:]

print("X：horizontal_feature，y：vertical_label")

print("training:(X_train:{}, y_train:{})\ntesting:(X_test:{},y_test:{})"
    .format(len(X_train),len(y_train),len(X_test),len(y_test)))
print("X：horizontal_feature，y：vertical_label")

training:(X_train:40, y_train:40)

testing:(X_test:10,y_test:10)

3. Visualization

import torch
from torch import nn
import matplotlib.pyplot as plt

def plot_prediction(train_data=X_train,
                     train_label=y_train,
                     test_data=X_test,
                     test_label=y_test,
                     prediction=None):
  """
  plot training data,test data and compare prediction
  """
  plt.figure(figsize=(10, 7))

  # plot training data in blue
  plt.scatter(train_data,train_label,c="b",s=4,label="Training data")

  # plot test data in green
  plt.scatter(test_data,test_label,c="g",s=4,label="Testing data")

  if prediction is not None:
    # plot prediction in red (prediction were made on test data)
    plt.scatter(test_data,prediction,c="r",s=4,label="Prediction")

  # show legend
  plt.legend(prop={"size": 14});

plot_prediction();

plt.savefig("VisualizeDataSplitting.svg")
VisualizeDataSplitting