Does Pandas Iterrows Have Performance Issues

Does pandas iterrows have performance issues?

Generally, iterrows should only be used in very, very specific cases. This is the general order of precedence for performance of various operations:

1) vectorization
2) using a custom cython routine
3) apply
    a) reductions that can be performed in cython
    b) iteration in python space
4) itertuples
5) iterrows
6) updating an empty frame (e.g. using loc one-row-at-a-time)

Using a custom Cython routine is usually too complicated, so let's skip that for now.

1) Vectorization is ALWAYS, ALWAYS the first and best choice. However, there is a small set of cases (usually involving a recurrence) which cannot be vectorized in obvious ways. Furthermore, on a smallish DataFrame, it may be faster to use other methods.

3) apply usually can be handled by an iterator in Cython space. This is handled internally by pandas, though it depends on what is going on inside the apply expression. For example, df.apply(lambda x: np.sum(x)) will be executed pretty swiftly, though of course, df.sum(1) is even better. However something like df.apply(lambda x: x['b'] + 1) will be executed in Python space, and consequently is much slower.

4) itertuples does not box the data into a Series. It just returns the data in the form of tuples.

5) iterrows DOES box the data into a Series. Unless you really need this, use another method.

6) Updating an empty frame a-single-row-at-a-time. I have seen this method used WAY too much. It is by far the slowest. It is probably common place (and reasonably fast for some python structures), but a DataFrame does a fair number of checks on indexing, so this will always be very slow to update a row at a time. Much better to create new structures and concat.

Performance issues with pandas iterrows

How does this fit you needs?

import pandas as pd

s = '''\
DOCNO   RSLTN1
MP00059189  72386.0
MP0059189A  65492.0
MP00066187  96497.0
MP00061663  43677.0
MP00063387  42465.0'''

# Recreate dataframe
df = pd.read_csv(pd.compat.StringIO(s), sep='\s+')

# Create mask 
# We sort to make sure we keep only highest value
# Remove all non-digit according to: https://stackoverflow.com/questions/44117326/
m = (df.sort_values(by='RSLTN1',ascending=False)['DOCNO']
       .str.extract('(\d+)', expand=False)
       .astype(int).duplicated())

# Apply inverted `~` mask 
df = df.loc[~m]

Resulting df:

        DOCNO   RSLTN1
0  MP00059189  72386.0
2  MP00066187  96497.0
3  MP00061663  43677.0
4  MP00063387  42465.0

In this example the following row was removed:

MP0059189A  65492.0

Iterrows performance

Assuming your empty cells are NaN values, this gives you the first non-NA value of each row for the group of columns you are interested in:

df[df>0][columns1].bfill(axis=1).iloc[:,0]

0     NaN
1     NaN
2     NaN
3     NaN
4     NaN
5    20.0
6     NaN
7    20.0
8     NaN

Thus, this will give you the abs(a-b) you're searching for:

res = (df[df>0][columns1].bfill(axis=1).iloc[:,0]
      -df[df>0][columns2].bfill(axis=1).iloc[:,0]).abs()
res

0        NaN
1        NaN
2        NaN
3        NaN
4        NaN
5    22977.5
6        NaN
7        NaN
8        NaN

You can either combine it with your initialized discount column:

res.combine_first(df.discount)

or fill the blanks:

res.fillna(0)

Iterating over rows in a dataframe in Pandas: is there a difference between using df.index and df.iterrows() as iterators?

When we doing for loop , look up index get the data require additional loc

for index in df.index:
    value = df.loc['index','col']

When we do df.iterrows

for index, row in df.iterrows():
    value = row['col']

Since you already with pandas , both of them are not recommended. Unless you need certain function and cannot be vectorized.

However, IMO, I preferred df.index

Speeding up loop over dataframes

I have faced a similar problem, using itertuples instead of iterrows shows significant reduction in time.
why iterrows have issues.
Hope this helps.

---

Related Topics