为什么 groupby 不传播过滤器上下文？

Question

我为我的数据框中的每个特征创建了一个列来计算 pct_change 并且它工作正常，这是代码：

features_to_measure_growth_for = ['Base','Total_NMV_x','Total_orders','Frequency','Retention_rate','Inorganic_Activations', 'Inorganic_NMV', 'Organic_Activations',
       'Organic_NMV', 'Visits', 'strike_rate', 'one_order', 'Two_orders',
       'Three_orders', 'Four_orders']

for feature in features_to_measure_growth_for:
    top_10_per_city[f"{feature} Growth %"] = top_10_per_city.groupby('polygon')[feature].pct_change()

print("Making PCT Change columns is done")

当我想创造复合增长时，我得到了所有特性中的所有

NaN

，这是代码：

def compound_growth(group, feature):
    end_val = group[group['period'] == group['period'].max()][feature].iloc[0]
    beg_val = group[group['period'] == group['period'].min()][feature].iloc[0]
    n_periods = group.shape[0] - 1
    if n_periods > 0:
        return ((end_val / beg_val) ** (1 / n_periods) - 1) * 100
    else:
        return 0.0


for feature in features_to_measure_growth_for:
    top_10_per_city[f"{feature} Compound Growth %"] = top_10_per_city.groupby('polygon').apply(compound_growth, feature = feature)

print('Making The Compoung Growth Rate is Done.')

我试过像这样在函数中打印每个变量的输出：

print(group['polygon'].iloc[0],feature,end_val, beg_val,n_periods)

输出：

6 Of October El 7osari 1 Base 51.0 43.0 11

6 October El 7osari 1 是多边形， 基础是特征， 51.0 是起始值， 43.0 是结束值， 11 是期数。

感谢您的帮助。