利用csv导入数据集，并用jupyter设计程序计算数据集的熵和信息增益

以下是利用Python的pandas库和numpy库导入csv数据集并计算熵和信息增益的程序。

首先，导入相关库：

import pandas as pd
import numpy as np
from math import log2

然后，读取csv数据集：

df = pd.read_csv('dataset.csv')

接下来，定义一个函数来计算数据集的熵：

def entropy(data):
    # 获取数据集的标签列
    labels = data.iloc[:, -1]
    # 统计每个标签出现的次数
    label_counts = labels.value_counts()
    # 计算熵
    entropy = 0
    for count in label_counts:
        p = count / len(labels)
        entropy -= p * log2(p)
    return entropy

然后，定义一个函数来计算数据集某一特征的信息增益：

def info_gain(data, feature):
    # 获取数据集的标签列
    labels = data.iloc[:, -1]
    # 获取特征列
    feature_col = data[feature]
    # 获取特征的取值列表
    feature_values = feature_col.unique()
    # 计算数据集的熵
    data_entropy = entropy(data)
    # 计算条件熵
    feature_entropy = 0
    for value in feature_values:
        subset = data[feature_col == value]
        p = len(subset) / len(data)
        feature_entropy += p * entropy(subset)
    # 计算信息增益
    info_gain = data_entropy - feature_entropy
    return info_gain

最后，调用函数计算数据集的熵和信息增益：

# 计算数据集的熵
data_entropy = entropy(df)
print('数据集的熵为：', data_entropy)

# 计算每个特征的信息增益
for feature in df.columns[:-1]:
    info_gain_value = info_gain(df, feature)
    print('特征', feature, '的信息增益为：', info_gain_value)

完整代码如下：

import pandas as pd
import numpy as np
from math import log2

def entropy(data):
    # 获取数据集的标签列
    labels = data.iloc[:, -1]
    # 统计每个标签出现的次数
    label_counts = labels.value_counts()
    # 计算熵
    entropy = 0
    for count in label_counts:
        p = count / len(labels)
        entropy -= p * log2(p)
    return entropy

def info_gain(data, feature):
    # 获取数据集的标签列
    labels = data.iloc[:, -1]
    # 获取特征列
    feature_col = data[feature]
    # 获取特征的取值列表
    feature_values = feature_col.unique()
    # 计算数据集的熵
    data_entropy = entropy(data)
    # 计算条件熵
    feature_entropy = 0
    for value in feature_values:
        subset = data[feature_col == value]
        p = len(subset) / len(data)
        feature_entropy += p * entropy(subset)
    # 计算信息增益
    info_gain = data_entropy - feature_entropy
    return info_gain

# 读取csv数据集
df = pd.read_csv('dataset.csv')

# 计算数据集的熵
data_entropy = entropy(df)
print('数据集的熵为：', data_entropy)

# 计算每个特征的信息增益
for feature in df.columns[:-1]:
    info_gain_value = info_gain(df, feature)
    print('特征', feature, '的信息增益为：', info_gain_value)

注意：在实际使用中，应该根据具体的数据集和需求，适当地修改程序。