I have a dataframe with multiple columns. I want to filter the dataframe based on two columns. For one column there is one condition and for the other column there are 3 conditions.

This code for the one condition in each column works fine:

filtered_df = df[(df['col1'] == 'cond1') & (df['col2'] == 'cond2)]

When I expand this to 3 conditions for the second column I get an empty dataframe as a result.

filtered_df = df[(df['col1'] == 'cond1') & (df['col2'] == 'cond2) & (df['col2'] == 'cond3) & (df['col2'] == 'cond4)]

How can I filter with more than 2 conditions?

CodePudding user response：

(df['col2'] == 'cond2') & (df['col2'] == 'cond3')

There couldn't be a value that both equal to cond2 and equal to cond3 at the same time. You may want |

Better is to use isin for a list of value

df['col2'].isin(['cond2', 'cond3'])

CodePudding user response：

filtered_df = df[(df['col1'] == 'cond1') & (df['col2'] == 'cond2) & (df['col2'] == 'cond3) & (df['col2'] == 'cond4)]

please be careful... in your code there are four condition that contradictory:

col2 can not be cond2 and cond3 and cond4 at same time

so you must use or:

filtered_df = df[(df['col1'] == 'cond1') & ((df['col2'] == 'cond2') | (df['col2'] == 'cond3') | (df['col2'] == 'cond4'))]

CodePudding user response：

It's correct but there is missing ' in your code, try this:

filtered_df = df.loc[(df['col1'] == 'cond1') & (df['col2'] == 'cond2') & (df['col2'] == 'cond3') & (df['col2'] == 'cond4')]

CodePudding user response：

You can use pandas.query.

filtered_df = df.query('(col2=="cond2" or col2=="cond3" or col2=="cond4") and (col1 == "cond1")')