README.md 14.6 KB
Newer Older
C
choijulie 已提交
1 2
# Linear Regression
Let us begin the tutorial with a classical problem called Linear Regression \[[1](#References)\]. In this chapter, we will train a model from a realistic dataset to predict home prices. Some important concepts in Machine Learning will be covered through this example.
Z
zhouxiao-coder 已提交
3

L
Luo Tao 已提交
4
The source code for this tutorial lives on [book/fit_a_line](https://github.com/PaddlePaddle/book/tree/develop/01.fit_a_line). For instructions on getting started with PaddlePaddle, see [PaddlePaddle installation guide](https://github.com/PaddlePaddle/book/blob/develop/README.md#running-the-book).
L
Luo Tao 已提交
5

C
choijulie 已提交
6 7
## Problem Setup
Suppose we have a dataset of $n$ real estate properties. These real estate properties will be referred to as *homes* in this chapter for clarity.
Z
zhouxiao-coder 已提交
8

C
choijulie 已提交
9
Each home is associated with $d$ attributes. The attributes describe characteristics such the number of rooms in the home, the number of schools or hospitals in the neighborhood, and the traffic condition nearby.
Z
zhouxiao-coder 已提交
10

C
choijulie 已提交
11
In our problem setup, the attribute $x_{i,j}$ denotes the $j$th characteristic of the $i$th home. In addition, $y_i$ denotes the price of the $i$th home. Our task is to predict $y_i$ given a set of attributes $\{x_{i,1}, ..., x_{i,d}\}$. We assume that the price of a home is a linear combination of all of its attributes, namely,
Z
zhouxiao-coder 已提交
12

C
choijulie 已提交
13
$$y_i = \omega_1x_{i,1} + \omega_2x_{i,2} + \ldots + \omega_dx_{i,d} + b,  i=1,\ldots,n$$
Z
zhouxiao-coder 已提交
14

C
choijulie 已提交
15 16 17 18
where $\vec{\omega}$ and $b$ are the model parameters we want to estimate. Once they are learned, we will be able to predict the price of a home, given the attributes associated with it. We call this model **Linear Regression**. In other words, we want to regress a value against several values linearly. In practice, a linear model is often too simplistic to capture the real relationships between the variables. Yet, because Linear Regression is easy to train and analyze, it has been applied to a large number of real problems. As a result, it is an important topic in many classic Statistical Learning and Machine Learning textbooks \[[2,3,4](#References)\].

## Results Demonstration
We first show the result of our model. The dataset [UCI Housing Data Set](https://archive.ics.uci.edu/ml/datasets/Housing) is used to train a linear model to predict the home prices in Boston. The figure below shows the predictions the model makes for some home prices. The $X$-axis represents the median value of the prices of simlilar homes within a bin, while the $Y$-axis represents the home value our linear model predicts. The dotted line represents points where $X=Y$. When reading the diagram, the more precise the model predicts, the closer the point is to the dotted line.
Z
zhouxiao-coder 已提交
19
<p align="center">
C
choijulie 已提交
20 21
    <img src = "image/predictions_en.png" width=400><br/>
    Figure 1. Predicted Value V.S. Actual Value
Z
zhouxiao-coder 已提交
22 23
</p>

C
choijulie 已提交
24
## Model Overview
25

C
choijulie 已提交
26
### Model Definition
27

C
choijulie 已提交
28
In the UCI Housing Data Set, there are 13 home attributes $\{x_{i,j}\}$ that are related to the median home price $y_i$, which we aim to predict. Thus, our model can be written as:
Z
zhouxiao-coder 已提交
29 30 31

$$\hat{Y} = \omega_1X_{1} + \omega_2X_{2} + \ldots + \omega_{13}X_{13} + b$$

C
choijulie 已提交
32
where $\hat{Y}$ is the predicted value used to differentiate from actual value $Y$. The model learns parameters $\omega_1, \ldots, \omega_{13}, b$, where the entries of $\vec{\omega}$ are **weights** and $b$ is **bias**.
Z
zhouxiao-coder 已提交
33

C
choijulie 已提交
34
Now we need an objective to optimize, so that the learned parameters can make $\hat{Y}$ as close to $Y$ as possible. Let's refer to the concept of [Loss Function (Cost Function)](https://en.wikipedia.org/wiki/Loss_function). A loss function must output a non-negative value, given any pair of the actual value $y_i$ and the predicted value $\hat{y_i}$. This value reflects the magnitutude of the model error.
Z
zhouxiao-coder 已提交
35

C
choijulie 已提交
36
For Linear Regression, the most common loss function is [Mean Square Error (MSE)](https://en.wikipedia.org/wiki/Mean_squared_error) which has the following form:
Z
zhouxiao-coder 已提交
37

38
$$MSE=\frac{1}{n}\sum_{i=1}^{n}{(\hat{Y_i}-Y_i)}^2$$
Z
zhouxiao-coder 已提交
39

C
choijulie 已提交
40 41 42
That is, for a dataset of size $n$, MSE is the average value of the the prediction sqaure errors.

### Training
Z
zhouxiao-coder 已提交
43

C
choijulie 已提交
44 45 46 47 48
After setting up our model, there are several major steps to go through to train it:
1. Initialize the parameters including the weights $\vec{\omega}$ and the bias $b$. For example, we can set their mean values as $0$s, and their standard deviations as $1$s.
2. Feedforward. Evaluate the network output and compute the corresponding loss.
3. [Backpropagate](https://en.wikipedia.org/wiki/Backpropagation) the errors. The errors will be propagated from the output layer back to the input layer, during which the model parameters will be updated with the corresponding errors.
4. Repeat steps 2~3, until the loss is below a predefined threshold or the maximum number of repeats is reached.
49

C
choijulie 已提交
50
## Dataset
Y
Yi Wang 已提交
51

C
choijulie 已提交
52
### Python Dataset Modules
53

C
choijulie 已提交
54
Our program starts with importing necessary packages:
55

56 57 58
```python
import paddle.v2 as paddle
import paddle.v2.dataset.uci_housing as uci_housing
Z
zhouxiao-coder 已提交
59 60
```

C
choijulie 已提交
61
We encapsulated the [UCI Housing Data Set](https://archive.ics.uci.edu/ml/datasets/Housing) in our Python module `uci_housing`.  This module can
62

C
choijulie 已提交
63 64
1. download the dataset to `~/.cache/paddle/dataset/uci_housing/housing.data`, if not yet, and
2.  [preprocesses](#preprocessing) the dataset.
65

C
choijulie 已提交
66
### An Introduction of the Dataset
67

C
choijulie 已提交
68
The UCI housing dataset has 506 instances. Each instance describes the attributes of a house in surburban Boston.  The attributes are explained below:
69

C
choijulie 已提交
70
| Attribute Name | Characteristic | Data Type |
Z
zhouxiao-coder 已提交
71
| ------| ------ | ------ |
C
choijulie 已提交
72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101
| CRIM | per capita crime rate by town | Continuous|
| ZN | proportion of residential land zoned for lots over 25,000 sq.ft. | Continuous |
| INDUS | proportion of non-retail business acres per town | Continuous |
| CHAS | Charles River dummy variable | Discrete, 1 if tract bounds river; 0 otherwise|
| NOX | nitric oxides concentration (parts per 10 million) | Continuous |
| RM | average number of rooms per dwelling | Continuous |
| AGE | proportion of owner-occupied units built prior to 1940 | Continuous |
| DIS | weighted distances to five Boston employment centres | Continuous |
| RAD | index of accessibility to radial highways | Continuous |
| TAX | full-value property-tax rate per $10,000 | Continuous |
| PTRATIO | pupil-teacher ratio by town | Continuous |
| B | 1000(Bk - 0.63)^2 where Bk is the proportion of blacks by town | Continuous |
| LSTAT | % lower status of the population | Continuous |
| MEDV | Median value of owner-occupied homes in $1000's | Continuous |

The last entry is the median home price.

### Preprocessing
#### Continuous and Discrete Data
We define a feature vector of length 13 for each home, where each entry corresponds to an attribute. Our first observation is that, among the 13 dimensions, there are 12 continuous dimensions and 1 discrete dimension.

Note that although a discrete value is also written as numeric values such as 0, 1, or 2, its meaning differs from a continuous value drastically.  The linear difference between two discrete values has no meaning. For example, suppose $0$, $1$, and $2$ are used to represent colors *Red*, *Green*, and *Blue* respectively. Judging from the numeric representation of these colors, *Red* differs more from *Blue* than it does from *Green*. Yet in actuality, it is not true that extent to which the color *Blue* is different from *Red* is greater than the extent to which *Green* is different from *Red*. Therefore, when handling a discrete feature that has $d$ possible values, we usually convert it to $d$ new features where each feature takes a binary value, $0$ or $1$, indicating whether the original value is absent or present. Alternatively, the discrete features can be mapped onto a continuous multi-dimensional vector through an embedding table. For our problem here, because CHAS itself is a binary discrete value, we do not need to do any preprocessing.

#### Feature Normalization
We also observe a huge difference among the value ranges of the 13 features (Figure 2). For instance, the values of feature *B* fall in $[0.32, 396.90]$, whereas those of feature *NOX* has a range of $[0.3850, 0.8170]$. An effective optimization would require data normalization. The goal of data normalization is to scale te values of each feature into roughly the same range, perhaps $[-0.5, 0.5]$. Here, we adopt a popular normalization technique where we substract the mean value from the feature value and divide the result by the width of the original range.

There are at least three reasons for [Feature Normalization](https://en.wikipedia.org/wiki/Feature_scaling) (Feature Scaling):
- A value range that is too large or too small might cause floating number overflow or underflow during computation.
- Different value ranges might result in varying *importances* of different features to the model (at least in the beginning of the training process). This assumption about the data is often unreasonable, making the optimization difficult, which in turn results in increased training time.
- Many machine learning techniques or models (e.g., *L1/L2 regularization* and *Vector Space Model*) assumes that all the features have roughly zero means and their value ranges are similar.
Z
zhouxiao-coder 已提交
102 103

<p align="center">
C
choijulie 已提交
104 105
    <img src = "image/ranges_en.png" width=550><br/>
    Figure 2. The value ranges of the features
Z
zhouxiao-coder 已提交
106 107
</p>

C
choijulie 已提交
108 109 110
#### Prepare Training and Test Sets
We split the dataset in two, one for adjusting the model parameters, namely, for model training, and the other for model testing. The model error on the former is called the **training error**, and the error on the latter is called the **test error**. Our goal in training a model is to find the statistical dependency between the outputs and the inputs, so that we can predict new outputs given new inputs. As a result, the test error reflects the performance of the model better than the training error does. We consider two things when deciding the ratio of the training set to the test set: 1) More training data will decrease the variance of the parameter estimation, yielding more reliable models; 2) More test data will decrease the variance of the test error, yielding more reliable test errors. One standard split ratio is $8:2$.

111

C
choijulie 已提交
112
When training complex models, we usually have one more split: the validation set. Complex models usually have [Hyperparameters](https://en.wikipedia.org/wiki/Hyperparameter_optimization) that need to be set before the training process, such as the number of layers in the network. Because hyperparameters are not part of the model parameters, they cannot be trained using the same loss function. Thus we will try several sets of hyperparameters to train several models and cross-validate them on the validation set to pick the best one; finally, the selected trained model is tested on the test set. Because our model is relatively simple, we will omit this validation process.
113

Z
zhouxiao-coder 已提交
114

C
choijulie 已提交
115
## Training
Z
zhouxiao-coder 已提交
116

C
choijulie 已提交
117
`fit_a_line/trainer.py` demonstrates the training using [PaddlePaddle](http://paddlepaddle.org).
Y
Yi Wang 已提交
118

C
choijulie 已提交
119
### Initialize PaddlePaddle
Z
zhouxiao-coder 已提交
120

121
```python
122
paddle.init(use_gpu=False, trainer_count=1)
Z
zhouxiao-coder 已提交
123 124
```

C
choijulie 已提交
125
### Model Configuration
Z
zhouxiao-coder 已提交
126

C
choijulie 已提交
127
Logistic regression is essentially a fully-connected layer with linear activation:
Z
zhouxiao-coder 已提交
128

129
```python
130 131
x = paddle.layer.data(name='x', type=paddle.data_type.dense_vector(13))
y_predict = paddle.layer.fc(input=x,
G
gongweibao 已提交
132
                                size=1,
133 134
                                act=paddle.activation.Linear())
y = paddle.layer.data(name='y', type=paddle.data_type.dense_vector(1))
L
Luo Tao 已提交
135
cost = paddle.layer.mse_cost(input=y_predict, label=y)
136
```
C
choijulie 已提交
137
### Create Parameters
Z
zhouxiao-coder 已提交
138

139
```python
140
parameters = paddle.parameters.create(cost)
Z
zhouxiao-coder 已提交
141 142
```

C
choijulie 已提交
143
### Create Trainer
144

Z
zhouxiao-coder 已提交
145
```python
G
gongweibao 已提交
146 147
optimizer = paddle.optimizer.Momentum(momentum=0)

148 149 150
trainer = paddle.trainer.SGD(cost=cost,
                             parameters=parameters,
                             update_equation=optimizer)
Z
zhouxiao-coder 已提交
151 152
```

C
choijulie 已提交
153
### Feeding Data
Y
Yi Wang 已提交
154

C
choijulie 已提交
155 156 157
PaddlePaddle provides the
[reader mechanism](https://github.com/PaddlePaddle/Paddle/tree/develop/doc/design/reader)
for loadinng training data. A reader may return multiple columns, and we need a Python dictionary to specify the mapping from column index to data layers.
158

Z
zhouxiao-coder 已提交
159
```python
Y
Yi Wang 已提交
160 161 162
feeding={'x': 0, 'y': 1}
```

C
choijulie 已提交
163
Moreover, an event handler is provided to print the training progress:
164

L
liaogang 已提交
165
```python
L
liaogang 已提交
166
# event_handler to print training and testing info
L
liaogang 已提交
167 168 169
def event_handler(event):
    if isinstance(event, paddle.event.EndIteration):
        if event.batch_id % 100 == 0:
L
liaogang 已提交
170 171 172
            print "Pass %d, Batch %d, Cost %f" % (
                event.pass_id, event.batch_id, event.cost)

L
liaogang 已提交
173
    if isinstance(event, paddle.event.EndPass):
L
liaogang 已提交
174 175 176 177 178
        result = trainer.test(
            reader=paddle.batch(
                uci_housing.test(), batch_size=2),
            feeding=feeding)
        print "Test %d, Cost %f" % (event.pass_id, result.cost)
L
liaogang 已提交
179 180
```

Y
Yi Wang 已提交
181
```python
182
# event_handler to print training and testing info
Q
qiaolongfei 已提交
183
from paddle.v2.plot import Ploter
Q
qiaolongfei 已提交
184

Q
qiaolongfei 已提交
185 186
train_title = "Train cost"
test_title = "Test cost"
C
choijulie 已提交
187
plot_cost = Ploter(train_title, test_title)
Q
qiaolongfei 已提交
188

Q
qiaolongfei 已提交
189
step = 0
Q
qiaolongfei 已提交
190

L
liaogang 已提交
191
def event_handler_plot(event):
Q
qiaolongfei 已提交
192
    global step
193
    if isinstance(event, paddle.event.EndIteration):
Q
qiaolongfei 已提交
194
        if step % 10 == 0:  # every 10 batches, record a train cost
C
choijulie 已提交
195
            plot_cost.append(train_title, step, event.cost)
Q
qiaolongfei 已提交
196

Q
qiaolongfei 已提交
197
        if step % 100 == 0: # every 100 batches, record a test cost
Q
qiaolongfei 已提交
198 199 200 201
            result = trainer.test(
                reader=paddle.batch(
                    uci_housing.test(), batch_size=2),
                feeding=feeding)
C
choijulie 已提交
202
            plot_cost.append(test_title, step, result.cost)
Q
qiaolongfei 已提交
203 204

        if step % 100 == 0: # every 100 batches, update cost plot
C
choijulie 已提交
205
            plot_cost.plot()
Q
qiaolongfei 已提交
206

Q
qiaolongfei 已提交
207
        step += 1
Z
zhouxiao-coder 已提交
208
```
Y
Yi Wang 已提交
209

C
choijulie 已提交
210
### Start Training
Z
zhouxiao-coder 已提交
211

212
```python
213
trainer.train(
G
gongweibao 已提交
214
    reader=paddle.batch(
215 216 217
        paddle.reader.shuffle(
            uci_housing.train(), buf_size=500),
        batch_size=2),
G
gongweibao 已提交
218
    feeding=feeding,
L
liaogang 已提交
219
    event_handler=event_handler_plot,
220
    num_passes=30)
Z
zhouxiao-coder 已提交
221 222
```

Q
qiaolongfei 已提交
223
![png](./image/train_and_test.png)
Q
qiaolongfei 已提交
224

C
choijulie 已提交
225 226
## Summary
This chapter introduces *Linear Regression* and how to train and test this model with PaddlePaddle, using the UCI Housing Data Set. Because a large number of more complex models and techniques are derived from linear regression, it is important to understand its underlying theory and limitation.
Z
zhouxiao-coder 已提交
227 228


C
choijulie 已提交
229
## References
Z
zhouxiao-coder 已提交
230 231 232
1. https://en.wikipedia.org/wiki/Linear_regression
2. Friedman J, Hastie T, Tibshirani R. The elements of statistical learning[M]. Springer, Berlin: Springer series in statistics, 2001.
3. Murphy K P. Machine learning: a probabilistic perspective[M]. MIT press, 2012.
Z
zhouxiao-coder 已提交
233
4. Bishop C M. Pattern recognition[J]. Machine Learning, 2006, 128.
L
Luo Tao 已提交
234 235

<br/>
L
Luo Tao 已提交
236
This tutorial is contributed by <a xmlns:cc="http://creativecommons.org/ns#" href="http://book.paddlepaddle.org" property="cc:attributionName" rel="cc:attributionURL">PaddlePaddle</a>, and licensed under a <a rel="license" href="http://creativecommons.org/licenses/by-sa/4.0/">Creative Commons Attribution-ShareAlike 4.0 International License</a>.