代码用途:
根据输入(test_tuple和params)的长度(可以是不同的长度),从主DataFrame(df)生成已过滤的DataFrame(filtered_df)。可能有数百种过滤器组合。
发布原因:
我想说的是,这里的一切都可以产生预期的输出。话虽这么说,我不喜欢解决方案1的实现方法,在该方法中,在dict中创建了一个虚拟DataFrame对象,该对象依次循环过滤和更新。解决方案1似乎有点模糊,但是我需要一些指导以实现更简洁的方法。
请求:
有没有办法使用解决方案2所示的filter_t?
filter_t的格式正确,但是是字符串。是否有产生filter_t的方法,以便可以如图所示使用它?
输入示例:
test_tuple = [('Serial Number', [12345]),
('Test Points', ['TestpointA', 'TestpointC']),
('Voltage_1', [3.0, 3.3, 3.6, 0.0]),
('Temperature Setpoint', [0, 60]),
('Slew_1', [200, 400, 800, 1600, 3200, 6400])]
params = ['sn', 'tp', 'v1', 'temp', 'slew']
代码:
for i in itertools.product(*[b for _, b in test_tuple]):
print('\n'.join(f'{a}:{b}' for a, b in zip(params, i)))
name_params = '_'.join(f'{b}{a}' for a, b in zip(params, i))
filter_t = ' & '.join(f'(self.df["{c[0]}"] == {b})' for b, c in zip(i, test_tuple))
print(f'filter_t: {filter_t}')
filter_l = [(c[0], b) for b, c in zip(i, test_tuple)]
filter_t的格式为字符串:
filter_t: (self.df["Test Points"] == 3P3V) & (self.df["Slew_1"] == 5000)
filter_t: (self.df["Serial Number"] == 2450) & (self.df["Test Points"] == 3P3V) & (self.df["Voltage_1"] == 11.6) & (self.df["Temperature Setpoint"] == 25.0) & (self.df["Slew_1"] == 5000)
filter_l的格式:
filter_l: [('Test Points', '3P3V_Edge'), ('Slew_1', 200)]
filter_l: [('Serial Number', 1234), ('Test Points', '3P3V'), ('Voltage_1', 11.6), ('Temperature Setpoint', 25.0), ('Slew_1', 200)]
解决方案1-正常运行:
filtered_df = {1: df}
for x in filter_l:
filtered_df[1] = (filtered_df[1].loc[(filtered_df[1][f'{x[0]}'] == x[1])])
解决方案2-可能吗?:
df_filter = self.df.loc[filter_t]
过滤器示例:
df_filter = self.df.loc[(self.df['Serial Number'] == 1234) &
(self.df['Test Points'] == '3P3V') &
(self.df['Voltage_1'] == 11.6) &
(self.df['Temperature Setpoint'] == 25.0) &
(self.df['Slew'] == 200)]
最终解决方案@John Zwinck-Thx:
filter_t = ' & '.join(f'{c[0]} == "{b}"' for b, c in zip(i, test_tuple))
filtered_df = df.loc[df.eval(filter_t)]
print(f'filter_t: {filter_t}')
>>> filter_t: Test_Points == "3P3V" & Slew_1 == "5000"
提前感谢您的建议。
答案 0 :(得分:1)
如果可以安装numexpr
,建议您尝试DataFrame.query()。
首先,您需要在列名称中用下划线(或什么都没有)替换空格。然后,构建并使用如下所示的过滤器字符串:
filter_str = 'Serial_Number == 2450 and Test_Points == "3P3V" and Voltage_1 == 11.6'
df_filter = df.query(filter_str)
如果您安装了numexpr
,这可能是最快的解决方案。
或者,构建一个蒙版列表,然后将其组合。这类似于但优于解决方案1:
masks = [df[x[0]] == x[1] for x in filter_l] # list of bool arrays
filtered_df = df[np.logical_and.reduce(masks)] # combine and apply masks