使用不带NaN的.lt(number).groupby().transform('all')

import pandas as pd

# 创建一个示例DataFrame
data = {
    'group': ['A', 'A', 'B', 'B', 'C', 'C'],
    'sales': [100, 150, 200, 250, 300, 350]
}
df = pd.DataFrame(data)

# 设置阈值
threshold = 200

# 使用不带NaN的.lt(number).groupby().transform('all')
result = df[df['sales'].lt(threshold)].groupby('group')['sales'].transform('all')

# 将结果添加到原始DataFrame中
df['all_below_threshold'] = result

print(df)

# 去除含有NaN值的行
df_clean = df.dropna(subset=['sales'])
result_clean = df_clean[df_clean['sales'].lt(threshold)].groupby('group')['sales'].transform('all')

# 或者填充NaN值
df_filled = df.fillna({'sales': 0})  # 假设0是一个合理的填充值
result_filled = df_filled[df_filled['sales'].lt(threshold)].groupby('group')['sales'].transform('all')

# 筛选出所有分组中所有值都小于阈值的分组
groups_below_threshold = df[result][df['group'] == df['group']].groupby('group').groups.keys()