CODEWITHSUNDEEP
pythonpandas
mttk
mttk

Reputation: 940

Extract value from single row of pandas DataFrame

I have a dataset in a relational database format (linked by ID's over various .csv files).

I know that each data frame contains only one value of an ID, and I'd like to know the simplest way to extract values from that row.

What I'm doing now:

# the group has only one element
purchase_group = purchase_groups.get_group(user_id)
price = list(purchase_group['Column_name'])[0]

The third row is bothering me as it seems ugly, however I'm not sure what is the workaround. The grouping (I guess) assumes that there might be multiple values and returns a <class 'pandas.core.frame.DataFrame'> object, while I'd like just a row returned.

Upvotes: 43

Views: 113425

Answers (5)

Ander Arana
Ander Arana

Reputation: 53

I've come into this article over time on many occasions. And it's the typical thing that I have written in different repositories in a non-consistent way. So I've done a little benchmark of which method is faster.

For the tests I have used a dataframe in which there is a single column and a single row.

df['_col0'].values[0] -> 3.34 µs ± 4.65 ns
df['_col0'].item() -> 4.7 µs ± 10.6 ns
df.values[0][0] -> 6.36 µs ± 611 ns
np.array(df)[0][0] -> 10.4 µs ± 1.19 µs
np.array(df['_col0'])[0] -> 11.2 µs ± 162 ns
df.squeeze() -> 18.3 µs ± 177 ns
df.iloc[0, 0] -> 18.8 µs ± 5.66 µs

Upvotes: 2

jesseWUT
jesseWUT

Reputation: 651

Late to the party here, but purchase_group['Column Name'].item() is now available and is cleaner than some other solutions

Upvotes: 14

dna-data
dna-data

Reputation: 93

This method is intuitive; for example to get the first row (list from a list of lists) of values from the dataframe:

np.array(df)[0]

Upvotes: 0

Yesh
Yesh

Reputation: 1194

If purchase_group has single row then doing purchase_group = purchase_group.squeeze() would make it into a series so you could simply call purchase_group['Column_name'] to get your values

Upvotes: 10

EdChum
EdChum

Reputation: 394459

If you want just the value and not a df/series then call values and index the first element [0] so just:

price = purchase_group['Column_name'].values[0]

will work.

Upvotes: 84

Related Questions