监督学习: 为CharityML寻找捐献者

开始

在这个项目中，你将使用1994年美国人口普查收集的数据，选用几个监督学习算法以准确地建模被调查者的收入。然后，你将根据初步结果从中选择出最佳的候选算法，并进一步优化该算法以最好地建模这些数据。你的目标是建立一个能够准确地预测被调查者年收入是否超过50000美元的模型。这种类型的任务会出现在那些依赖于捐款而存在的非营利性组织。了解人群的收入情况可以帮助一个非营利性的机构更好地了解他们要多大的捐赠，或是否他们应该接触这些人。虽然我们很难直接从公开的资源中推断出一个人的一般收入阶层，但是我们可以（也正是我们将要做的）从其他的一些公开的可获得的资源中获得一些特征从而推断出该值。

这个项目的数据集来自UCI机器学习知识库。这个数据集是由Ron Kohavi和Barry Becker在发表文章“Scaling Up the Accuracy of Naive-Bayes Classifiers: A Decision-Tree Hybrid”之后捐赠的，你可以在Ron Kohavi提供的在线版本中找到这个文章。我们在这里探索的数据集相比于原有的数据集有一些小小的改变，比如说移除了特征'fnlwgt' 以及一些遗失的或者是格式不正确的记录。

探索数据

运行下面的代码单元以载入需要的Python库并导入人口普查数据。注意数据集的最后一列'income'将是我们需要预测的列（表示被调查者的年收入会大于或者是最多50,000美元），人口普查数据中的每一列都将是关于被调查者的特征。

In [1]:

# 检查你的Python版本
from sys import version_info
if version_info.major != 2 and version_info.minor != 7:
    raise Exception('请使用Python 2.7来完成此项目')

In [5]:

# 为这个项目导入需要的库
import numpy as np
import pandas as pd
from time import time
from IPython.display import display # 允许为DataFrame使用display()# 导入附加的可视化代码visuals.py
import visuals as vs# 为notebook提供更加漂亮的可视化
%matplotlib inline# 导入人口普查数据
data = pd.read_csv("census.csv")# 成功 - 显示第一条记录
display(data.head(n=1))

练习：数据探索

首先我们对数据集进行一个粗略的探索，我们将看看每一个类别里会有多少被调查者？并且告诉我们这些里面多大比例是年收入大于50,000美元的。在下面的代码单元中，你将需要计算以下量：

• 总的记录数量，'n_records'
• 年收入大于50,000美元的人数，'n_greater_50k'.
• 年收入最多为50,000美元的人数 'n_at_most_50k'.
• 年收入大于50,000美元的人所占的比例， 'greater_percent'.

提示： 您可能需要查看上面的生成的表，以了解'income'条目的格式是什么样的。

In [6]:

# TODO：总的记录数
n_records = len(data)# TODO：被调查者的收入大于$50,000的人数
n_greater_50k = len(data[data['income']=='>50K'])# TODO：被调查者的收入最多为$50,000的人数
n_at_most_50k = len(data[data['income']=='<=50K'])# TODO：被调查者收入大于$50,000所占的比例greater_percent = float(n_greater_50k)/n_records*100# 打印结果
print "Total number of records: {}".format(n_records)
print "Individuals making more than $50,000: {}".format(n_greater_50k)
print "Individuals making at most $50,000: {}".format(n_at_most_50k)
print "Percentage of individuals making more than $50,000: {:.2f}%".format(greater_percent)

Total number of records: 45222
Individuals making more than $50,000: 11208
Individuals making at most $50,000: 34014
Percentage of individuals making more than $50,000: 24.78%

准备数据

在数据能够被作为输入提供给机器学习算法之前，它经常需要被清洗，格式化，和重新组织 – 这通常被叫做预处理。幸运的是，对于这个数据集，没有我们必须处理的无效或丢失的条目，然而，由于某一些特征存在的特性我们必须进行一定的调整。这个预处理都可以极大地帮助我们提升几乎所有的学习算法的结果和预测能力。

获得特征和标签

income 列是我们需要的标签，记录一个人的年收入是否高于50K。 因此我们应该把他从数据中剥离出来，单独存放。

In [7]:

# 将数据切分成特征和对应的标签
income_raw = data['income']
#print income_raw
features_raw = data.drop('income', axis = 1)#drop _ delete,operation:axis = 1  column ,axis = 0 ,row
#print features_raw

转换倾斜的连续特征

一个数据集有时可能包含至少一个靠近某个数字的特征，但有时也会有一些相对来说存在极大值或者极小值的不平凡分布的的特征。算法对这种分布的数据会十分敏感，并且如果这种数据没有能够很好地规一化处理会使得算法表现不佳。在人口普查数据集的两个特征符合这个描述：’capital-gain'和'capital-loss'。

运行下面的代码单元以创建一个关于这两个特征的条形图。请注意当前的值的范围和它们是如何分布的。

In [8]:

# 可视化 'capital-gain'和'capital-loss' 两个特征
vs.distribution(features_raw)

 aaarticlea/png;base64,iVBORw0KGgoAAAANSUhEUgAAAxAAAAF2CAYAAAD+y36TAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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%0A2gQEAADQJiAAAIA2AQEAALQJCAAAoO0Ga7UEP7pgqRcAAAAASUVORK5CYII=” alt=”” /> 

对于高度倾斜分布的特征如'capital-gain'和'capital-loss'，常见的做法是对数据施加一个对数转换，将数据转换成对数，这样非常大和非常小的值不会对学习算法产生负面的影响。并且使用对数变换显著降低了由于异常值所造成的数据范围异常。但是在应用这个变换时必须小心：因为0的对数是没有定义的，所以我们必须先将数据处理成一个比0稍微大一点的数以成功完成对数转换。

运行下面的代码单元来执行数据的转换和可视化结果。再次，注意值的范围和它们是如何分布的。

In [9]:

# 对于倾斜的数据使用Log转换
skewed = ['capital-gain', 'capital-loss']
features_raw[skewed] = data[skewed].apply(lambda x: np.log(x + 1))  # x=log(x+1)# 可视化对数转换后 'capital-gain'和'capital-loss' 两个特征
vs.distribution(features_raw, transformed = True)

 aaarticlea/png;base64,iVBORw0KGgoAAAANSUhEUgAAAxAAAAF2CAYAAAD+y36TAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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%0Alt95AAAAAElFTkSuQmCC” alt=”” /> 

规一化数字特征

除了对于高度倾斜的特征施加转换，对数值特征施加一些形式的缩放通常会是一个好的习惯。在数据上面施加一个缩放并不会改变数据分布的形式（比如上面说的’capital-gain’ or ‘capital-loss’）；但是，规一化保证了每一个特征在使用监督学习器的时候能够被平等的对待。注意一旦使用了缩放，观察数据的原始形式不再具有它本来的意义了，就像下面的例子展示的。

运行下面的代码单元来规一化每一个数字特征。我们将使用sklearn.preprocessing.MinMaxScaler来完成这个任务。

In [10]:

from sklearn.preprocessing import MinMaxScaler# 初始化一个 scaler，并将它施加到特征上
scaler = MinMaxScaler()#range = (0,1)
numerical = ['age', 'education-num', 'capital-gain', 'capital-loss', 'hours-per-week']
features_raw[numerical] = scaler.fit_transform(data[numerical])# 显示一个经过缩放的样例记录
display(features_raw.head(n = 1))
#display(data.head(n=1))

练习：（非数字）数据预处理

从上面的数据探索中的表中，我们可以看到有几个属性的每一条记录都是非数字的。通常情况下，学习算法期望输入是数字的，这要求非数字的特征（称为类别变量）被转换。转换类别变量的一种流行的方法是使用独热编码方案。独热编码为每一个非数字特征的每一个可能的类别创建一个“虚拟”变量。例如，假设someFeature有三个可能的取值A，B或者C，。我们将把这个特征编码成someFeature_A, someFeature_B和someFeature_C.

此外，对于非数字的特征，我们需要将非数字的标签'income'转换成数值以保证学习算法能够正常工作。因为这个标签只有两种可能的类别（”<=50K”和”>50K”），我们不必要使用独热编码，可以直接将他们编码分别成两个类0和1，在下面的代码单元中你将实现以下功能：

• 使用pandas.get_dummies()对'features_raw'数据来施加一个独热编码。
• 将目标标签'income_raw'转换成数字项。
  • 将”<=50K”转换成0；将”>50K”转换成1。

In [11]:

# TODO：使用pandas.get_dummies()对'features_raw'数据进行独热编码
features = pd.get_dummies(features_raw)# TODO：将'income_raw'编码成数字值
income = income_raw.apply(lambda x:0 if (x=='<=50K') else 1)
# Python 里用 lambda 来定义匿名函数
# 匿名函数lambda x: x * x实际上就是：
#   def f(x):
#    return x * x
# 打印经过独热编码之后的特征数量
encoded = list(features.columns)
print "{} total features after one-hot encoding.".format(len(encoded))# 移除下面一行的注释以观察编码的特征名字print encoded

103 total features after one-hot encoding.
['age', 'education-num', 'capital-gain', 'capital-loss', 'hours-per-week', 'workclass_ Federal-gov', 'workclass_ Local-gov', 'workclass_ Private', 'workclass_ Self-emp-inc', 'workclass_ Self-emp-not-inc', 'workclass_ State-gov', 'workclass_ Without-pay', 'education_level_ 10th', 'education_level_ 11th', 'education_level_ 12th', 'education_level_ 1st-4th', 'education_level_ 5th-6th', 'education_level_ 7th-8th', 'education_level_ 9th', 'education_level_ Assoc-acdm', 'education_level_ Assoc-voc', 'education_level_ Bachelors', 'education_level_ Doctorate', 'education_level_ HS-grad', 'education_level_ Masters', 'education_level_ Preschool', 'education_level_ Prof-school', 'education_level_ Some-college', 'marital-status_ Divorced', 'marital-status_ Married-AF-spouse', 'marital-status_ Married-civ-spouse', 'marital-status_ Married-spouse-absent', 'marital-status_ Never-married', 'marital-status_ Separated', 'marital-status_ Widowed', 'occupation_ Adm-clerical', 'occupation_ Armed-Forces', 'occupation_ Craft-repair', 'occupation_ Exec-managerial', 'occupation_ Farming-fishing', 'occupation_ Handlers-cleaners', 'occupation_ Machine-op-inspct', 'occupation_ Other-service', 'occupation_ Priv-house-serv', 'occupation_ Prof-specialty', 'occupation_ Protective-serv', 'occupation_ Sales', 'occupation_ Tech-support', 'occupation_ Transport-moving', 'relationship_ Husband', 'relationship_ Not-in-family', 'relationship_ Other-relative', 'relationship_ Own-child', 'relationship_ Unmarried', 'relationship_ Wife', 'race_ Amer-Indian-Eskimo', 'race_ Asian-Pac-Islander', 'race_ Black', 'race_ Other', 'race_ White', 'sex_ Female', 'sex_ Male', 'native-country_ Cambodia', 'native-country_ Canada', 'native-country_ China', 'native-country_ Columbia', 'native-country_ Cuba', 'native-country_ Dominican-Republic', 'native-country_ Ecuador', 'native-country_ El-Salvador', 'native-country_ England', 'native-country_ France', 'native-country_ Germany', 'native-country_ Greece', 'native-country_ Guatemala', 'native-country_ Haiti', 'native-country_ Holand-Netherlands', 'native-country_ Honduras', 'native-country_ Hong', 'native-country_ Hungary', 'native-country_ India', 'native-country_ Iran', 'native-country_ Ireland', 'native-country_ Italy', 'native-country_ Jamaica', 'native-country_ Japan', 'native-country_ Laos', 'native-country_ Mexico', 'native-country_ Nicaragua', 'native-country_ Outlying-US(Guam-USVI-etc)', 'native-country_ Peru', 'native-country_ Philippines', 'native-country_ Poland', 'native-country_ Portugal', 'native-country_ Puerto-Rico', 'native-country_ Scotland', 'native-country_ South', 'native-country_ Taiwan', 'native-country_ Thailand', 'native-country_ Trinadad&Tobago', 'native-country_ United-States', 'native-country_ Vietnam', 'native-country_ Yugoslavia']

混洗和切分数据

现在所有的 类别变量 已被转换成数值特征，而且所有的数值特征已被规一化。和我们一般情况下做的一样，我们现在将数据（包括特征和它们的标签）切分成训练和测试集。其中80%的数据将用于训练和20%的数据用于测试。然后再进一步把训练数据分为训练集和验证集，用来选择和优化模型。

运行下面的代码单元来完成切分。

In [12]:

# 导入 train_test_split
from sklearn.model_selection import train_test_split# 将'features'和'income'数据切分成训练集和测试集
X_train, X_test, y_train, y_test = train_test_split(features, income, test_size = 0.2, random_state = 0,
                                                    stratify = income)
# 将'X_train'和'y_train'进一步切分为训练集和验证集
X_train, X_val, y_train, y_val = train_test_split(X_train, y_train, test_size=0.2, random_state=0,
                                                    stratify = y_train)# 显示切分的结果
print "Training set has {} samples.".format(X_train.shape[0])
print "Validation set has {} samples.".format(X_val.shape[0])
print "Testing set has {} samples.".format(X_test.shape[0])

Training set has 28941 samples.
Validation set has 7236 samples.
Testing set has 9045 samples.

评价模型性能

在这一部分中，我们将尝试四种不同的算法，并确定哪一个能够最好地建模数据。四种算法包含一个天真的预测器 和三个你选择的监督学习器。

评价方法和朴素的预测器

CharityML通过他们的研究人员知道被调查者的年收入大于$50,000最有可能向他们捐款。因为这个原因*CharityML*对于准确预测谁能够获得$50,000以上收入尤其有兴趣。这样看起来使用准确率作为评价模型的标准是合适的。另外，把没有收入大于$50,000的人识别成年收入大于$50,000对于CharityML来说是有害的，因为他想要找到的是有意愿捐款的用户。这样，我们期望的模型具有准确预测那些能够年收入大于$50,000的能力比模型去查全这些被调查者更重要。我们能够使用F-beta score作为评价指标，这样能够同时考虑查准率和查全率：

Fβ=(1+β2)⋅precision⋅recall(β2⋅precision)+recallFβ=(1+β2)⋅precision⋅recall(β2⋅precision)+recall

尤其是，当 β=0.5β=0.5 的时候更多的强调查准率，这叫做F0.50.5 score （或者为了简单叫做F-score）。

问题 1 – 天真的预测器的性能

通过查看收入超过和不超过 $50,000 的人数，我们能发现多数被调查者年收入没有超过 $50,000。如果我们简单地预测说“这个人的收入没有超过 $50,000”，我们就可以得到一个 准确率超过 50% 的预测。这样我们甚至不用看数据就能做到一个准确率超过 50%。这样一个预测被称作是天真的。通常对数据使用一个天真的预测器是十分重要的，这样能够帮助建立一个模型表现是否好的基准。 使用下面的代码单元计算天真的预测器的相关性能。将你的计算结果赋值给'accuracy', ‘precision’, ‘recall’ 和 'fscore'，这些值会在后面被使用，请注意这里不能使用scikit-learn，你需要根据公式自己实现相关计算。

如果我们选择一个无论什么情况都预测被调查者年收入大于 $50,000 的模型，那么这个模型在验证集上的准确率，查准率，查全率和 F-score是多少？

In [14]:

#不能使用scikit-learn，你需要根据公式自己实现相关计算。# #TODO： 计算准确率
# accuracy = float(n_greater_50k)/n_records# # TODO： 计算查准率 Precision
# precision = float(n_greater_50k)/n_records# # TODO： 计算查全率 Recall
# recall = float(n_greater_50k)/n_greater_50k# # TODO： 使用上面的公式，设置beta=0.5，计算F-score
# fscore = (1+0.5**2)*(precision*recall)/((0.5**2*precision)+recall)TP = float(len(y_val[y_val == 1]))
FP = float(len(y_val[y_val == 0]))
FN = 0
#TODO： 计算准确率
accuracy = TF / len(y_val)# TODO： 计算查准率 Precision
precision = TP / (TP+FP)# TODO： 计算查全率 Recall
recall = TP / (TP+FN)# TODO： 使用上面的公式，设置beta=0.5，计算F-score
fscore = (1+0.5**2)* ( (precision*recall) / (0.5**2*precision + recall) )# 打印结果
print "Naive Predictor on validation data: \n \
    Accuracy score: {:.4f} \n \
    Precision: {:.4f} \n \
    Recall: {:.4f} \n \
    F-score: {:.4f}".format(accuracy, precision, recall, fscore)

Naive Predictor on validation data:
     Accuracy score: 0.2478
     Precision: 0.2478
     Recall: 1.0000
     F-score: 0.2917

参考链接

监督学习模型

问题 2 – 模型应用

你能够在 scikit-learn 中选择以下监督学习模型

• 高斯朴素贝叶斯 (GaussianNB)
• 决策树 (DecisionTree)
• 集成方法 (Bagging, AdaBoost, Random Forest, Gradient Boosting)
• K近邻 (K Nearest Neighbors)
• 随机梯度下降分类器 (SGDC)
• 支撑向量机 (SVM)
• Logistic回归（LogisticRegression）

从上面的监督学习模型中选择三个适合我们这个问题的模型，并回答相应问题。

模型1

模型名称

回答：支撑向量机 (SVM)

描述一个该模型在真实世界的一个应用场景。（你需要为此做点研究，并给出你的引用出处）

回答：利用SVM进行文本分类 对文本自动做SVM模型的训练。包括Libsvm、Liblinear包的选择，分词，词典生成，特征选择，SVM参数的选优，SVM模型的训练等都可以一步完成。示意图如下

引用：文本分类与SVM

这个模型的优势是什么？他什么情况下表现最好？

回答： 优点： 1）SVM 是一种小样本学习方法。不涉及大数定律等,不同于现有的统计方法。

从本质上看,它避开了从归纳到演绎的传统过程,实现了高效的从训练样本到预报样本的“转导推理”,大大简化了通常的分类和回归等问题。

2）SVM 最终决策函数只由少数的支持向量确定,计算复杂性不取决于样本空间的维数,而取决于支持向量的数目。一定程度上避免了“维数灾难”。

3）什么情况下表现最好？噪声较小，小样本。

这个模型的缺点是什么？什么条件下它表现很差？

回答：

(1) 不适合大规模训练样本

SVM通过二次规划求解支持向量，二次规划的求解涉及n阶矩阵计算（n为样本个数）。大矩阵计算和存储将耗费大量计算时间。

(2) 难以解决多分类问题

经典SVM算法只适合二分类问题。而实际应用中常出现多分类问题。 可通过构造多个分类器的组合改善这一问题。

参考链接：SVM的特点与不足

根据我们当前数据集的特点，为什么这个模型适合这个问题。

回答：数据集没有过大，支持向量机适合处理二分类问题

模型2

模型名称

回答：高斯朴素贝叶斯 (GaussianNB)

描述一个该模型在真实世界的一个应用场景。（你需要为此做点研究，并给出你的引用出处）

回答：过滤垃圾邮件 可以把文档中的词作为特征进行分类

参考链接

这个模型的优势是什么？他什么情况下表现最好？

回答：对缺失数据不敏感，适合小规模数据

这个模型的缺点是什么？什么条件下它表现很差？

回答：1)对输入数据的形式敏感

2)朴素贝叶斯模型假设各属性相互独立。但在实际应用中，属性之间往往有一定关联性，导致分类效果受到影响。

根据我们当前数据集的特点，为什么这个模型适合这个问题。

回答：数据集各属性关联性相对较小，且为小规模数据

模型3

模型名称

回答：集成方法 (Bagging, AdaBoost, Random Forest, Gradient Boosting)

描述一个该模型在真实世界的一个应用场景。（你需要为此做点研究，并给出你的引用出处）

回答：声纳信号分类——Random Forest

使用神经网络的声纳信号分类中使用的数据集。任务是训练一个模型来区分声纳信号。 引用

这个模型的优势是什么？他什么情况下表现最好？

回答：1）不易发生过拟合

2）准确率很高

这个模型的缺点是什么？什么条件下它表现很差？

回答：1）对异常值敏感

2）若数据分布不均匀，可能导致分类精度下降

根据我们当前数据集的特点，为什么这个模型适合这个问题。

回答：本数据集数据分布较均匀，异常值较少

练习 – 创建一个训练和预测的流水线

为了正确评估你选择的每一个模型的性能，创建一个能够帮助你快速有效地使用不同大小的训练集并在验证集上做预测的训练和验证的流水线是十分重要的。 你在这里实现的功能将会在接下来的部分中被用到。在下面的代码单元中，你将实现以下功能：

• 从sklearn.metrics中导入fbeta_score和accuracy_score。
• 用训练集拟合学习器，并记录训练时间。
• 对训练集的前300个数据点和验证集进行预测并记录预测时间。
• 计算预测训练集的前300个数据点的准确率和F-score。
• 计算预测验证集的准确率和F-score。

In [11]:

# TODO：从sklearn中导入两个评价指标 - fbeta_score和accuracy_score
from sklearn.metrics import fbeta_score, accuracy_scoredef train_predict(learner, sample_size, X_train, y_train, X_val, y_val):
    '''
    inputs:
       - learner: the learning algorithm to be trained and predicted on
       - sample_size: the size of samples (number) to be drawn from training set
       - X_train: features training set
       - y_train: income training set
       - X_val: features validation set
       - y_val: income validation set
    '''
    #
    #fbeta_score:计算F值,即召回率与准确率的加权调和平均,
#     该函数在多标签分类(一个样本有多个标签)时有用,如果只使用准确率作为度量,
#     模型只要把所有输入分类为"所有类别"就可以获得完美的准确率,
#     为了避免这种情况,度量指标应该对错误的选择进行惩罚.
#     F-beta分值(0到1之间)通过准确率和召回率的加权调和平均来更好的度量.
#     当beta为1时,该指标等价于F-measure,beta<1时,模型选对正确的标签更加重要,
#     而beta>1时,模型对选错标签有更大的惩罚.    results = {}    # TODO：使用sample_size大小的训练数据来拟合学习器
    # TODO: Fit the learner to the training data using slicing with 'sample_size'
    start = time() # 获得程序开始时间
    learner = learner.fit(X_train[:sample_size],y_train[:sample_size])#训练算法
    end = time() # 获得程序结束时间    # TODO：计算训练时间
    results['train_time'] = end - start    # TODO: 得到在验证集上的预测值
    #       然后得到对前300个训练数据的预测结果
    start = time() # 获得程序开始时间
    predictions_val = learner.predict(X_val)    predictions_train = learner.predict(X_train[:300])
    end = time() # 获得程序结束时间    # TODO：计算预测用时
    results['pred_time'] = end - start    # TODO：计算在最前面的300个训练数据的准确率
    results['acc_train'] = accuracy_score(y_train[:300],predictions_train)    # TODO：计算在验证上的准确率
    results['acc_val'] = accuracy_score(y_val,predictions_val)    # TODO：计算在最前面300个训练数据上的F-score
    results['f_train'] = fbeta_score(y_train[:300],predictions_train,0.5)    # TODO：计算验证集上的F-score
    results['f_val'] = fbeta_score(y_val,predictions_val,0.5)    # 成功
    print "{} trained on {} samples.".format(learner.__class__.__name__, sample_size)    # 返回结果
    return results

练习：初始模型的评估

在下面的代码单元中，您将需要实现以下功能：

• 导入你在前面讨论的三个监督学习模型。
• 初始化三个模型并存储在'clf_A'，'clf_B'和'clf_C'中。
  • 使用模型的默认参数值，在接下来的部分中你将需要对某一个模型的参数进行调整。
  • 设置random_state (如果有这个参数)。
• 计算1%， 10%， 100%的训练数据分别对应多少个数据点，并将这些值存储在'samples_1', 'samples_10', 'samples_100'中

注意：取决于你选择的算法，下面实现的代码可能需要一些时间来运行！

In [13]:

# TODO：从sklearn中导入三个监督学习模型
# svm 贝叶斯、集成方法
from sklearn.svm import SVC
from sklearn.naive_bayes import GaussianNB
from sklearn.ensemble import AdaBoostClassifier# TODO：初始化三个模型
clf_A = SVC(random_state=6)
clf_B = GaussianNB()
clf_C = AdaBoostClassifier(random_state=6)# TODO：计算1%， 10%， 100%的训练数据分别对应多少点
samples_1 = int(len(X_train)*0.01)
samples_10 = int(len(X_train)*0.10)
samples_100 = int(len(X_train))# 收集学习器的结果
results = {}
for clf in [clf_A, clf_B, clf_C]:
    clf_name = clf.__class__.__name__
    results[clf_name] = {}
    for i, samples in enumerate([samples_1, samples_10, samples_100]):
        results[clf_name][i] = train_predict(clf, samples, X_train, y_train, X_val, y_val)# 对选择的三个模型得到的评价结果进行可视化
vs.evaluate(results, accuracy, fscore)

SVC trained on 289 samples.
SVC trained on 2894 samples.
SVC trained on 28941 samples.
GaussianNB trained on 289 samples.
GaussianNB trained on 2894 samples.
GaussianNB trained on 28941 samples.
AdaBoostClassifier trained on 289 samples.
AdaBoostClassifier trained on 2894 samples.
AdaBoostClassifier trained on 28941 samples.

 aaarticlea/png;base64,iVBORw0KGgoAAAANSUhEUgAAAxAAAAIuCAYAAAAv/u6UAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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%0AdiAkSZIk9ez/ATLLKXhGz7w0AAAAAElFTkSuQmCC” alt=”” /> 

提高效果

在这最后一节中，您将从三个有监督的学习模型中选择 最好的 模型来使用学生数据。你将在整个训练集（X_train和y_train）上使用网格搜索优化至少调节一个参数以获得一个比没有调节之前更好的 F-score。

问题 3 – 选择最佳的模型

基于你前面做的评价，用一到两段话向 CharityML 解释这三个模型中哪一个对于判断被调查者的年收入大于 $50,000 是最合适的。
提示：你的答案应该包括评价指标，预测/训练时间，以及该算法是否适合这里的数据。

aaarticlea/png;base64,iVBORw0KGgoAAAANSUhEUgAAAxQAAAIuCAYAAAAmFU7uAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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%0A/wMylbbti3xsGgAAAABJRU5ErkJggg==” alt=”image.png” />

回答：

首选模型：集成方法之AdaBoostClassifier

理由1-预测/训练时间

无论是对于此样本的大数据集还是小数据集，AdaBoostClassifier和GaussianNB的预测/训练时间均相对较短， 而SVC模型的预测/训练时间随数据集增加而显著增加

理由2-accuracy score 准确率

在验证集上，AdaBoostClassifier和SVC模型的预测准确率相差不大，而GaussianNB的准确率则难以满足要求。

理由3-F-score

AdaBoostClassifier和SVC模型的F-score相差不大，而GaussianNB的F-score则难以满足要求。

综上所述，首选模型为AdaBoostClassifier

问题 4 – 用通俗的话解释模型

用一到两段话，向 CharityML 用外行也听得懂的话来解释最终模型是如何工作的。你需要解释所选模型的主要特点。例如，这个模型是怎样被训练的，它又是如何做出预测的。避免使用高级的数学或技术术语，不要使用公式或特定的算法名词。

回答：

集成方法（ensemble method）通过多个基分类器（base classifier）组合来完成模型训练。有点像谚语“三个臭皮匠顶个诸葛亮”。基分类器一般采用的是弱可学习（weakly learnable）分类器，通过集成方法，组合成一个强可学习（strongly learnable）分类器。

所谓弱可学习，是指学习的正确率仅略优于随机猜测的多项式学习算法；强可学习指正确率较高的多项式学习算法。

集成学习的泛化能力一般比单一基分类器情况要好。 Adaboost采用迭代的思想，每次迭代只训练一个弱分类器，训练好的弱分类器将参与下一次迭代的使用。

1）它的弱学习器是怎样的？

理论上可以选择任何一个分类或者回归学习器，不过需要支持样本权重。我们常用的一般是CART决策树或者神经网络MLP。默认是决策树，即AdaBoostClassifier默认使用CART分类树DecisionTreeClassifier，而AdaBoostRegressor默认使用CART回归树DecisionTreeRegressor。另外有一个要注意的点是，如果我们选择的AdaBoostClassifier算法是SAMME.R，则我们的弱分类学习器还需要支持概率预测，也就是在scikit-learn中弱分类学习器对应的预测方法除了predict还需要有predict_proba。

2）训练过程如何？如何更新样本权重？

为了构造出一个强的学习算法，首先需要选定一个弱学习算法，并利用同一个训练集不断训练弱学习算法，以提升弱学习算法的性能。在AdaBoost算法中，有两个权重，第一个数训练集中每个样本有一个权重，称为样本权重，用向量D表示；另一个是每一个弱学习算法具有一个权重，用向量α表示。假设有N个样本的训练集 {(X1,y1),(X2,y2),⋯,(Xn,yn)}，初始时，设定每个样本的权重是相等的，即1n，利用第一个弱学习算法h1对其进行学习，学习完成后进行错误率ε的统计：

ε=errorallε=errorall

其中，error表示被错误分类的样本数目，all表示所有样本的数目。这样便可以利用错误率ε计算弱学习算法的权重α1：

α1=12ln(1−εε)α1=12ln(1−εε)

在第一次学习完成后，需要重新调整样本的权重，以使得在第一分类中被错分的样本的权重，使得在接下来的学习中可以重点对其进行学习：

其中，表示ht(xi)=yi对第i个样本训练正确，ht(xi)≠yi表示对第i个样本训练错误。是一个Zt归一化因子：Zt=sum(D)

这样进行第二次的学习，当学习t轮后，得到了t个弱学习算法h1,⋯,ht及其权重α1,⋯,αt。对新的分类数据，分别计算t个弱分类器的输出h1(X),⋯,ht(X)，最终的AdaBoost算法的输出结果为：

H(X)=sign(∑i=1tαihi(X))H(X)=sign(∑i=1tαihi(X))

3）如何综合弱学习器的结果进行预测？

adaBoost分类器利用同一种基分类器（弱分类器），基于分类器的错误率分配不同的权重参数，最后累加加权的预测结果作为输出

参考链接：

https://www.cnblogs.com/Neo007/p/8460052.html

http://scikit-learn.org/stable/modules/generated/sklearn.ensemble.AdaBoostClassifier.html

https://blog.csdn.net/SzM21C11U68n04vdcLmJ/article/details/78348933?locationNum=3&fps=1

https://blog.csdn.net/qq_24753293/article/details/80180944

https://blog.csdn.net/px_528/article/details/72963977

练习：模型调优

调节选择的模型的参数。使用网格搜索（GridSearchCV）来至少调整模型的重要参数（至少调整一个），这个参数至少需尝试3个不同的值。你要使用整个训练集来完成这个过程。在接下来的代码单元中，你需要实现以下功能：

• 导入sklearn.model_selection.GridSearchCV 和 sklearn.metrics.make_scorer.
• 初始化你选择的分类器，并将其存储在clf中。
  • 设置random_state (如果有这个参数)。
• 创建一个对于这个模型你希望调整参数的字典。
  • 例如: parameters = {‘parameter’ : [list of values]}。
  • 注意： 如果你的学习器有 max_features 参数，请不要调节它！
• 使用make_scorer来创建一个fbeta_score评分对象（设置β=0.5β=0.5）。
• 在分类器clf上用’scorer’作为评价函数运行网格搜索，并将结果存储在grid_obj中。
• 用训练集（X_train, y_train）训练grid search object,并将结果存储在grid_fit中。

注意： 取决于你选择的参数列表，下面实现的代码可能需要花一些时间运行！

In [20]:

# TODO：导入'GridSearchCV', 'make_scorer'和其他一些需要的库
from sklearn.model_selection import GridSearchCV,KFold
from sklearn.metrics import make_scorer
from sklearn.ensemble import AdaBoostClassifier
# TODO：初始化分类器
clf = AdaBoostClassifier(random_state=6)# TODO：创建你希望调节的参数列表
# type：dictionary(key-value)
parameters = {'learning_rate':[1,5,10],'n_estimators': [25, 50, 100]}# TODO：创建一个fbeta_score打分对象
scorer = make_scorer(fbeta_score,beta=2)# TODO：在分类器上使用网格搜索，使用'scorer'作为评价函数grid_obj = GridSearchCV(estimator=clf,param_grid=parameters,scoring=scorer,cv=KFold(n_splits=12))# TODO：用训练数据拟合网格搜索对象并找到最佳参数grid_obj_fit = grid_obj.fit(X_train,y_train)# 得到estimator
best_clf = grid_obj.best_estimator_# 使用没有调优的模型做预测
predictions = (clf.fit(X_train, y_train)).predict(X_val)
best_predictions = best_clf.predict(X_val)# 汇报调参前和调参后的分数
print "Unoptimized model\n------"
print "Accuracy score on validation data: {:.4f}".format(accuracy_score(y_val, predictions))
print "F-score on validation data: {:.4f}".format(fbeta_score(y_val, predictions, beta = 0.5))
print "\nOptimized Model\n------"
print "Final accuracy score on the validation data: {:.4f}".format(accuracy_score(y_val, best_predictions))
print "Final F-score on the validation data: {:.4f}".format(fbeta_score(y_val, best_predictions, beta = 0.5))

Unoptimized model
------
Accuracy score on validation data: 0.8648
F-score on validation data: 0.7443Optimized Model
------
Final accuracy score on the validation data: 0.8684
Final F-score on the validation data: 0.7517

问题 5 – 最终模型评估

你的最优模型在测试数据上的准确率和 F-score 是多少？这些分数比没有优化的模型好还是差？你优化的结果相比于你在问题 1中得到的天真预测器怎么样？
注意：请在下面的表格中填写你的结果，然后在答案框中提供讨论。

结果:

回答：

一开始只调整了learning_rate参数，发现优化效果不明显。 后查阅资料发现实际调参时需要同时考虑n_estimators 以及learning_rate两个参数 很明显，不管优化与否，AdaBoost模型明显好于天真预测器。

n_estimators： 表示弱学习器的最大迭代次数，或者说最大的弱学习器的个数。一般来说n_estimators太小，容易欠拟合，n_estimators太大，又容易过拟合，一般选择一个适中的数值。默认是50。

learning_rate: 表示每个弱学习器的权重缩减系数ν，加上了正则化项，强学习器的迭代公式为fk(x)=fk−1(x)+ναkGk(x)。ν的取值范围为0<ν≤1。对于同样的训练集拟合效果，较小的ν意味着我们需要更多的弱学习器的迭代次数。通常我们用步长和迭代最大次数一起来决定算法的拟合效果。所以这两个参数n_estimators和learning_rate要一起调参。一般来说，可以从一个小一点的ν开始调参，默认是1。

参考链接

特征的重要性

在数据上（比如我们这里使用的人口普查的数据）使用监督学习算法的一个重要的任务是决定哪些特征能够提供最强的预测能力。专注于少量的有效特征和标签之间的关系，我们能够更加简单地理解这些现象，这在很多情况下都是十分有用的。在这个项目的情境下这表示我们希望选择一小部分特征，这些特征能够在预测被调查者是否年收入大于$50,000这个问题上有很强的预测能力。

选择一个有 'feature_importance_' 属性的scikit学习分类器（例如 AdaBoost，随机森林）。'feature_importance_' 属性是对特征的重要性排序的函数。在下一个代码单元中用这个分类器拟合训练集数据并使用这个属性来决定人口普查数据中最重要的5个特征。

问题 6 – 观察特征相关性

当探索数据的时候，它显示在这个人口普查数据集中每一条记录我们有十三个可用的特征。
在这十三个记录中，你认为哪五个特征对于预测是最重要的，选择每个特征的理由是什么？你会怎样对他们排序？

回答：

• 特征1:education_level 通常来看，受教育程度与平均收入水平息息相关。受教育程度较高的群体相对收入较高
• 特征2:occupation 不同职业领域的收入平均水平不同
• 特征3:workclass 被调查者的通常劳动类型与收入有关，比如without-pay 群体显然收入不会很高
• 特征4:hours-per-week 收入与工作时长有一定关系
• 特征5:relationship 家庭成员情况也能在一定情况下反映收入水平。

练习 – 提取特征重要性

选择一个scikit-learn中有feature_importance_属性的监督学习分类器，这个属性是一个在做预测的时候根据所选择的算法来对特征重要性进行排序的功能。

在下面的代码单元中，你将要实现以下功能：

• 如果这个模型和你前面使用的三个模型不一样的话从sklearn中导入一个监督学习模型。
• 在整个训练集上训练一个监督学习模型。
• 使用模型中的 'feature_importances_'提取特征的重要性。

In [19]:

# TODO：导入一个有'feature_importances_'的监督学习模型#已导入AdaBoostClassifier# TODO：在训练集上训练一个监督学习模型
model = AdaBoostClassifier(random_state =5)
model.fit(X_train,y_train)# TODO： 提取特征重要性
importances = model.feature_importances_# 绘图
vs.feature_plot(importances, X_train, y_train)

 aaarticlea/png;base64,iVBORw0KGgoAAAANSUhEUgAAAoAAAAFgCAYAAAArYcg8AAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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%0AZtYwDgDNzMzMGub/VGuhCtz6yMYAAAAASUVORK5CYII=” alt=”” /> 

问题 7 – 提取特征重要性

观察上面创建的展示五个用于预测被调查者年收入是否大于$50,000最相关的特征的可视化图像。

这五个特征的权重加起来是否超过了0.5?
这五个特征和你在问题 6中讨论的特征比较怎么样？
如果说你的答案和这里的相近，那么这个可视化怎样佐证了你的想法？
如果你的选择不相近，那么为什么你觉得这些特征更加相关？

回答：

1.五个特征权重之和0.56，超过了0.5

2.和我在题6中主观预测的特征比较差别较大，可见有时候主观判断是不可靠的，应该相信数据

3.capital-gain和capital-loss文档中没有说明现实意义

但是其他三个特征可以理解

1）年龄越大，相对来说资历越深的人收入水平越高

2）hours-per-week 每周工作时间一定程度上影响收入，高收入者平均加班时间较长

3）Education-num 可以反映受教育程度，受教育程度越高，收入平均来说也更高。

特征选择

如果我们只是用可用特征的一个子集的话模型表现会怎么样？通过使用更少的特征来训练，在评价指标的角度来看我们的期望是训练和预测的时间会更少。从上面的可视化来看，我们可以看到前五个最重要的特征贡献了数据中所有特征中超过一半的重要性。这提示我们可以尝试去减小特征空间，简化模型需要学习的信息。下面代码单元将使用你前面发现的优化模型，并只使用五个最重要的特征在相同的训练集上训练模型。

In [16]:

# 导入克隆模型的功能
from sklearn.base import clone# 减小特征空间
X_train_reduced = X_train[X_train.columns.values[(np.argsort(importances)[::-1])[:5]]]
X_val_reduced = X_val[X_val.columns.values[(np.argsort(importances)[::-1])[:5]]]# 在前面的网格搜索的基础上训练一个“最好的”模型
clf_on_reduced = (clone(best_clf)).fit(X_train_reduced, y_train)# 做一个新的预测
reduced_predictions = clf_on_reduced.predict(X_val_reduced)# 对于每一个版本的数据汇报最终模型的分数
print "Final Model trained on full data\n------"
print "Accuracy on validation data: {:.4f}".format(accuracy_score(y_val, best_predictions))
print "F-score on validation data: {:.4f}".format(fbeta_score(y_val, best_predictions, beta = 0.5))
print "\nFinal Model trained on reduced data\n------"
print "Accuracy on validation data: {:.4f}".format(accuracy_score(y_val, reduced_predictions))
print "F-score on validation data: {:.4f}".format(fbeta_score(y_val, reduced_predictions, beta = 0.5))

Final Model trained on full data
------
Accuracy on validation data: 0.8684
F-score on validation data: 0.7517Final Model trained on reduced data
------
Accuracy on validation data: 0.8378
F-score on validation data: 0.6989

问题 8 – 特征选择的影响

最终模型在只是用五个特征的数据上和使用所有的特征数据上的 F-score 和准确率相比怎么样？
如果训练时间是一个要考虑的因素，你会考虑使用部分特征的数据作为你的训练集吗？

回答：

1.最终模型在reduced data 和full data 中的 F-score 和准确率相比相差不大

2.会。因为可以在对模型效果影响较小的情况下，大幅缩短训练时间。

问题 9 – 在测试集上测试你的模型

终于到了测试的时候，记住，测试集只能用一次。

使用你最有信心的模型，在测试集上测试，计算出准确率和 F-score。 简述你选择这个模型的原因，并分析测试结果

In [17]:

#TODO test your model on testing data and report accuracy and F scorefrom sklearn.metrics import fbeta_score,accuracy_score
test_predicitions = best_clf.predict(X_test)print "Predict on test data \n"
print "Accuracy score on test data: {:.4f}".format(accuracy_score(y_test, test_predicitions))
print "F-score on test data: {:.4f}".format(fbeta_score(y_test, test_predicitions, beta = 0.5))

Predict on test data Accuracy score on test data: 0.8626
F-score on test data: 0.7411

回答：

Accuracy score on test data: 0.8626 F-score on test data: 0.7411

AdaBoostClassifier模型训练时间相对SVM较短，且模型且保持了较高的Accuracy score和F-score