In a data inspection context, we typically filter rows by column values to get a sense of the data e.g. to check for surprising values. We cover the basics of filtering here and in detail in Filtering.
We wish to extract rows that meet a certain condition on column values. In this example, we filter out rows where the column col_1 has a value of 0.
df.loc[df['col_1'] == 0]
Here is how this works:
loc[] to the data frame df..loc[] for filtering rows by a boolean condition on column values.col_1 with the integer 0 and return only the rows where the condition evaluates to TRUE.We wish to extract rows that meet a certain condition on column values. We wish to return only a specified set of columns (and not all columns). This is useful while developing a filter to look at just the relevant columns to verify if the filtering works as expected.
In this example, we wish to return rows where the value of col_1 is greater than 0 and to return columns col_1 and col_3.
df.loc[df['col_1'] > 0, ['col_1','col_3']]
Here is how this works:
.loc[] is the indexer we use for both boolean filtering and for column selection by name, we can combine both operations in a single call to .loc[].df['col_1'] > 0 to the first argument of loc[] and the column selection logic ['col_1','col_3'] to the second argument (see Selecting by Name).