基础知识：

（待补充）

作业相关基本代码

import graphlab
#设置线程限制，节省内存，防止程序崩溃
graphlab.set_runtime_config("GRAPHLAB_DEFAULT_NUM_PYLAMBDA_WORKERS", 8)
#导入数据
products = graphlab.SFrame("amazon_baby.gl/")
#统计评论中的词语
products["word_count"] = graphlab.text_analytics.count_words(products["review"])
#选择12个词作为情感分析的输入
selected_words = ['awesome', 'great', 'fantastic', 'amazing', 'love', 'horrible', 'bad', 'terrible', 'awful', 'wow', 'hate']
#统计每一记录中各个词语的频数
def selected_words_count(dict):
    value = 0
    if i in dict:
        value = dict[i]
    return value
for i in selected_words:
    products[i] = products['word_count'] .apply(selected_words_count)   
#查看统计结果
products.head()

效果如下：

Paste_Image.png

#统计12个词出现的次数
def count(word):
    value = 0
    for i in products["word_count"]:
        if word in i:
            value += i.get(word)
    return value
list = []
for x in selected_words:
    list.append(count(x))
new_dict = dict(zip(selected_words,list))
print new_dict

输出如下结果：
{'fantastic': 932, 'love': 42065, 'bad': 3724, 'awesome': 2090, 'great': 45206, 'terrible': 748, 'amazing': 1363, 'horrible': 734, 'awful': 383, 'hate': 1220, 'wow': 144}

#去除评论中性的词
products = products[products['rating'] != 3]
#添加新列'sentiment'
products['sentiment'] = products['rating'] >= 4

训练情感分类器

#将数据分为train（80%）和test（20%）两部分
train_data,test_data = products.random_split(.8, seed=0)
selected_words_model=graphlab.logistic_classifier.create(train_data,
                                                     target='sentiment',
                                                     features=selected_words,
                                                     validation_set=test_data)

输出如下结果：
Logistic regression:

Number of examples : 133448
Number of classes : 2
Number of feature columns : 11
Number of unpacked features : 11
Number of coefficients : 12
Starting Newton Method

+-----------+----------+--------------+-------------------+---------------------+
| Iteration | Passes | Elapsed Time | Training-accuracy | Validation-accuracy |
+-----------+----------+--------------+-------------------+---------------------+
| 1 | 2 | 1.285266 | 0.844299 | 0.842842 |
| 2 | 3 | 1.485671 | 0.844186 | 0.842842 |
| 3 | 4 | 1.685881 | 0.844276 | 0.843142 |
| 4 | 5 | 1.896848 | 0.844269 | 0.843142 |
| 5 | 6 | 2.098513 | 0.844269 | 0.843142 |
| 6 | 7 | 2.300982 | 0.844269 | 0.843142 |
+-----------+----------+--------------+-------------------+---------------------+
SUCCESS: Optimal solution found.

检查模型系数

selected_words_model['coefficients'].print_rows(num_rows=12)

输出结果如下：

Paste_Image.png

评估模型

graphlab.canvas.set_target('ipynb')
selected_words_model.show(view='Evaluation')

selected_words_model.evaluate(test_data)

输出结果如下：

Paste_Image.png

Paste_Image.png

{'accuracy': 0.8431419649291376,
'auc': 0.6648096413721418,
'confusion_matrix': Columns:
target_label int
predicted_label int
count int

Rows: 4

Data:
+--------------+-----------------+-------+
| target_label | predicted_label | count |
+--------------+-----------------+-------+
| 0 | 0 | 234 |
| 0 | 1 | 5094 |
| 1 | 1 | 27846 |
| 1 | 0 | 130 |
+--------------+-----------------+-------+
[4 rows x 3 columns],
'f1_score': 0.914242563530107,
'log_loss': 0.4054747110366022,
'precision': 0.8453551912568306,
'recall': 0.9953531598513011,
'roc_curve': Columns:
threshold float
fpr float
tpr float
p int
n int

Rows: 100001

Data:
+-----------+-----+-----+-------+------+
| threshold | fpr | tpr | p | n |
+-----------+-----+-----+-------+------+
| 0.0 | 1.0 | 1.0 | 27976 | 5328 |
| 1e-05 | 1.0 | 1.0 | 27976 | 5328 |
| 2e-05 | 1.0 | 1.0 | 27976 | 5328 |
| 3e-05 | 1.0 | 1.0 | 27976 | 5328 |
| 4e-05 | 1.0 | 1.0 | 27976 | 5328 |
| 5e-05 | 1.0 | 1.0 | 27976 | 5328 |
| 6e-05 | 1.0 | 1.0 | 27976 | 5328 |
| 7e-05 | 1.0 | 1.0 | 27976 | 5328 |
| 8e-05 | 1.0 | 1.0 | 27976 | 5328 |
| 9e-05 | 1.0 | 1.0 | 27976 | 5328 |
+-----------+-----+-----+-------+------+
[100001 rows x 5 columns]
Note: Only the head of the SFrame is printed.
You can use print_rows(num_rows=m, num_columns=n) to print more rows and columns.}

应用模型进行预测（习题10）

diaper_champ_reviews= products[products['name'] == 'Baby Trend Diaper Champ']
selected_words_model.predict(diaper_champ_reviews[0:1], output_type='probability')

输出结果如下：
dtype: float
Rows: 1
[0.796940851290673]

使用所有词的模型进

sentiment_model = graphlab.logistic_classifier.create(train_data,
                                                     target='sentiment',
                                                     features=['word_count'],
                                                     validation_set=test_data)
sentiment_model.evaluate(test_data)

输出结果如下：
{'accuracy': 0.916256305548883,
'auc': 0.9446492867438502,
'confusion_matrix': Columns:
target_label int
predicted_label int
count int

Rows: 4

Data:
+--------------+-----------------+-------+
| target_label | predicted_label | count |
+--------------+-----------------+-------+
| 0 | 1 | 1328 |
| 0 | 0 | 4000 |
| 1 | 1 | 26515 |
| 1 | 0 | 1461 |
+--------------+-----------------+-------+
[4 rows x 3 columns],
'f1_score': 0.9500349343413533,
'log_loss': 0.26106698432422365,
'precision': 0.9523039902309378,
'recall': 0.9477766657134686,
'roc_curve': Columns:
threshold float
fpr float
tpr float
p int
n int

Rows: 100001

Data:
+-----------+----------------+----------------+-------+------+
| threshold | fpr | tpr | p | n |
+-----------+----------------+----------------+-------+------+
| 0.0 | 1.0 | 1.0 | 27976 | 5328 |
| 1e-05 | 0.909346846847 | 0.998856162425 | 27976 | 5328 |
| 2e-05 | 0.896021021021 | 0.998748927652 | 27976 | 5328 |
| 3e-05 | 0.886448948949 | 0.998462968259 | 27976 | 5328 |
| 4e-05 | 0.879692192192 | 0.998284243637 | 27976 | 5328 |
| 5e-05 | 0.875187687688 | 0.998212753789 | 27976 | 5328 |
| 6e-05 | 0.872184684685 | 0.998177008865 | 27976 | 5328 |
| 7e-05 | 0.868618618619 | 0.998034029168 | 27976 | 5328 |
| 8e-05 | 0.864677177177 | 0.997998284244 | 27976 | 5328 |
| 9e-05 | 0.860735735736 | 0.997962539319 | 27976 | 5328 |
+-----------+----------------+----------------+-------+------+
[100001 rows x 5 columns]
Note: Only the head of the SFrame is printed.
You can use print_rows(num_rows=m, num_columns=n) to print more rows and columns.}

习题9

diaper_champ_reviews= products[products['name'] == 'Baby Trend Diaper Champ']
diaper_champ_reviews['predicted_sentiment'] = sentiment_model.predict(diaper_champ_reviews, output_type='probability')
diaper_champ_reviews =diaper_champ_reviews .sort('predicted_sentiment', ascending=False)
diaper_champ_reviews[0:1]

输出结果如下：

Paste_Image.png

Quiz Week 3

1. Out of the 11 words in selected_words, which one is most used in the reviews in the dataset?
   -great

2. Out of the 11 words in selected_words, which one is least used in the reviews in the dataset?
   -wow

3. Out of the 11 words in selected_words, which one got the most positive weight in the selected_words_model?
   -love

4. Out of the 11 words in selected_words, which one got the most negative weight in the selected_words_model?
   -terrible

5. Which of the following ranges contains the accuracy of the selected_words_model on the test_data?
   -0.843

6. Which of the following ranges contains the accuracy of the sentiment_model in the IPython Notebook from lecture on the test_data?
   -.916

7. Which of the following ranges contains the accuracy of the majority class classifier, which simply predicts the majority class on the test_data?
   -.835

8. How do you compare the different learned models with the baseline approach where we are just predicting the majority class?
   -all words better and other almost same

9. Which of the following ranges contains the "predicted_sentiment" for the most positive review for "Baby Trend Diaper Champ", according to the sentiment_model from the IPython Notebook from lecture?
   -0.999999937267

10. Consider the most positive review for "Baby Trend Diaper Champ" according to the sentiment_model from the IPython Notebook from lecture. Which of the following ranges contains the predicted_sentiment for this review, if we use the selected_words_model to analyze it?
    -0.79694

11. Why is the value of the predicted_sentiment for the most positive review found using the sentiment_model much more positive than the value predicted using the selected_words_model?
    -None of the selected_words appeared in the text of this review.