创建一个函数，该函数将接受 DataFrame 作为输入并返回所有适当分类特征的饼图

Question

我可以使用“Churn”列创建 1 个饼图来对数据进行分组，但是，不确定如何创建一个函数来接受 DataFrame 作为输入并返回所有适当的分类特征和饼图。在饼图中显示百分比分布？

作为 DataFrame，我使用“电信客户流失.csv"

f,axes=plt.subplots(1,2,figsize=(17,7))
df_churn['Churn'].value_counts().plot.pie(autopct='%1.1f%%',ax=axes[0])
sns.countplot('Churn',data=df_churn,ax=axes[1])
axes[0].set_title('Categorical Variable Pie Chart')
plt.show()

Answer 1

我做了这样的事情，不确定我是否做对了：-

#%% PlotMultiplePie

输入：df = Pandas dataframe，categorical_features = 功能列表，dropna = 使用 NaN 的布尔变量

输出：打印多个 px.pie()

def PlotMultiplePie(df_churn,categorical_features = None,dropna = False):
# 设置 30 个唯一变量的阈值，超过 50 个会导致饼图难看
阈值 = 40

# if user did not set categorical_features 
if categorical_features == None: 
    categorical_features = df_churn.select_dtypes(['object','category']).columns.to_list()
    print(categorical_features)

# loop through the list of categorical_features 
for cat_feature in categorical_features: 
    num_unique = df_churn[cat_feature].nunique(dropna = dropna)
    num_missing = df_churn[cat_feature].isna().sum()
    # prints pie chart and info if unique values below threshold 
    if num_unique <= threshold:
        print('Pie Chart for: ', cat_feature)
        print('Number of Unique Values: ', num_unique)
        print('Number of Missing Values: ', num_missing)
        fig = px.pie(df_churn[cat_feature].value_counts(dropna = dropna), values=cat_feature, 
             names = df_churn[cat_feature].value_counts(dropna = dropna).index,title = cat_feature,template='ggplot2')
        fig.show()
    else: 
        print('Pie Chart for ',cat_feature,' is unavailable due high number of Unique Values ')
        print('Number of Unique Values: ', num_unique)
        print('Number of Missing Values: ', num_missing)
        print('\n')

Answer 2

这对我有用。定义了一个函数来绘制 dataframe 中所有分类变量的饼图。

#Function to plot Pie-Charts for all categorical variables in the dataframe
def pie_charts_for_CategoricalVar(df_pie,m):
    '''Takes in a dataframe(df_pie) and plots pie charts for all categorical columns. m = number of columns required in grid'''
    
    #get all the column names in the dataframe
    a = []
    for i in df_pie:
        a.append(i)
    
    #isolate the categorical variable names from a to b
    b = []
    for i in a:
        if (df[i].dtype.name) == 'category':
            b.append(i)
        
    plt.figure(figsize=(15, 12))
    plt.subplots_adjust(hspace=0.2)
    plt.suptitle("Pie-Charts for Categorical Variables in the dataframe", fontsize=18, y=0.95)
    
    # number of columns, as inputted while calling the function
    ncols = m
    # calculate number of rows
    nrows = len(b) // ncols + (len(b) % ncols > 0)
    
    # loop through the length of 'b' and keep track of index
    for n, i in enumerate(b):
        # add a new subplot iteratively using nrows and ncols
        ax = plt.subplot(nrows, ncols, n + 1)

        # filter df and plot 'i' on the new subplot axis
        df.groupby(i).size().plot(kind='pie', autopct='%.2f%%',ax=ax)
        
        ax.set_title(i.upper())
        ax.set_xlabel("")
        ax.set_ylabel("")
    plt.show()

#calling the function to plot pie-charts for categorical variable

pie_charts_for_CategoricalVar(df,5)   #dataframe, no. of cols in the grid