我有以下代码:
x1 = df[(df['event_name'] == 'Listen') & (df['cell'] == 'CONTROL')].yes_user_count
y1 = df[(df['event_name'] == 'Listen') & (df['cell'] != 'CONTROL')].yes_user_count
x_n = df[(df['event_name'] == 'Listen') & (df['cell'] == 'CONTROL')].total_user_count
y_n = df[(df['event_name'] == 'Listen') & (df['cell'] != 'CONTROL')].total_user_count
zscore, pval = proportions_ztest([x1, y1], [x_n, y_n])
但是，ztest功能不起作用。原因看起来是 x1、y1、x_n 和 y_n 是 panda 系列。我尝试使用 x1.astype(int) 但它似乎没有解决问题。
有人可以建议什么是最好的解决方案吗？
最佳答案
你想要.tolist():
x1 = df[(df['event_name'] == 'Listen') & (df['cell'] == 'CONTROL')].yes_user_count.tolist()
y1 = df[(df['event_name'] == 'Listen') & (df['cell'] != 'CONTROL')].yes_user_count.tolist()
x_n = df[(df['event_name'] == 'Listen') & (df['cell'] == 'CONTROL')].total_user_count.tolist()
y_n = df[(df['event_name'] == 'Listen') & (df['cell'] != 'CONTROL')].total_user_count.tolist()
zscore, pval = proportions_ztest([x1, y1], [x_n, y_n])
但我也建议使用 loc 并将列名称作为第二个值:
x1 = df.loc[(df['event_name'] == 'Listen') & (df['cell'] == 'CONTROL'), 'yes_user_count'].tolist()
y1 = df.loc[(df['event_name'] == 'Listen') & (df['cell'] != 'CONTROL'), 'yes_user_count'].tolist()
x_n = df.loc[(df['event_name'] == 'Listen') & (df['cell'] == 'CONTROL'), 'total_user_count'].tolist()
y_n = df.loc[(df['event_name'] == 'Listen') & (df['cell'] != 'CONTROL'), 'total_user_count'].tolist()
zscore, pval = proportions_ztest([x1, y1], [x_n, y_n])
或者更好:
mask1 = (df['event_name'] == 'Listen') & (df['cell'] == 'CONTROL')
mask2 = (df['event_name'] == 'Listen') & (df['cell'] != 'CONTROL')
x1 = df.loc[mask1, 'yes_user_count'].tolist()
y1 = df.loc[mask2, 'yes_user_count'].tolist()
x_n = df.loc[mask1, 'total_user_count'].tolist()
y_n = df.loc[mask2, 'total_user_count'].tolist()
zscore, pval = proportions_ztest([x1, y1], [x_n, y_n])
关于python - 将 panda 系列更改为 int，我们在Stack Overflow上找到一个类似的问题： https://stackoverflow.com/questions/55368946/