High-Performance Pandas: eval and query#
As we’ve already seen in previous chapters, the power of the PyData stack is built upon the ability of NumPy and Pandas to push basic operations into lower-level compiled code via an intuitive higher-level syntax: examples are vectorized/broadcasted operations in NumPy, and grouping-type operations in Pandas. While these abstractions are efficient and effective for many common use cases, they often rely on the creation of temporary intermediate objects, which can cause undue overhead in computational time and memory use.
To address this, Pandas includes some methods that allow you to directly access C-speed operations without costly allocation of intermediate arrays: eval and query, which rely on the NumExpr package.
In this chapter I will walk you through their use and give some rules of thumb about when you might think about using them.
Motivating query and eval: Compound Expressions#
We’ve seen previously that NumPy and Pandas support fast vectorized operations; for example, when adding the elements of two arrays:
import numpy as np
rng = np.random.default_rng(42)
x = rng.random(1000000)
y = rng.random(1000000)
%timeit x + y
2.21 ms ± 142 µs per loop (mean ± std. dev. of 7 runs, 100 loops each)
As discussed in Computation on NumPy Arrays: Universal Functions, this is much faster than doing the addition via a Python loop or comprehension:
%timeit np.fromiter((xi + yi for xi, yi in zip(x, y)),
dtype=x.dtype, count=len(x))
263 ms ± 43.4 ms per loop (mean ± std. dev. of 7 runs, 1 loop each)
But this abstraction can become less efficient when computing compound expressions. For example, consider the following expression:
mask = (x > 0.5) & (y < 0.5)
Because NumPy evaluates each subexpression, this is roughly equivalent to the following:
tmp1 = (x > 0.5)
tmp2 = (y < 0.5)
mask = tmp1 & tmp2
In other words, every intermediate step is explicitly allocated in memory. If the x and y arrays are very large, this can lead to significant memory and computational overhead.
The NumExpr library gives you the ability to compute this type of compound expression element by element, without the need to allocate full intermediate arrays.
The NumExpr documentation has more details, but for the time being it is sufficient to say that the library accepts a string giving the NumPy-style expression you’d like to compute:
import numexpr
mask_numexpr = numexpr.evaluate('(x > 0.5) & (y < 0.5)')
np.all(mask == mask_numexpr)
True
The benefit here is that NumExpr evaluates the expression in a way that avoids temporary arrays where possible, and thus can be much more efficient than NumPy, especially for long sequences of computations on large arrays.
The Pandas eval and query tools that we will discuss here are conceptually similar, and are essentially Pandas-specific wrappers of NumExpr functionality.
pandas.eval for Efficient Operations#
The eval function in Pandas uses string expressions to efficiently compute operations on DataFrame objects.
For example, consider the following data:
import pandas as pd
nrows, ncols = 100000, 100
df1, df2, df3, df4 = (pd.DataFrame(rng.random((nrows, ncols)))
for i in range(4))
To compute the sum of all four DataFrames using the typical Pandas approach, we can just write the sum:
%timeit df1 + df2 + df3 + df4
73.2 ms ± 6.72 ms per loop (mean ± std. dev. of 7 runs, 10 loops each)
The same result can be computed via pd.eval by constructing the expression as a string:
%timeit pd.eval('df1 + df2 + df3 + df4')
34 ms ± 4.2 ms per loop (mean ± std. dev. of 7 runs, 10 loops each)
The eval version of this expression is about 50% faster (and uses much less memory), while giving the same result:
np.allclose(df1 + df2 + df3 + df4,
pd.eval('df1 + df2 + df3 + df4'))
True
pd.eval supports a wide range of operations.
To demonstrate these, we’ll use the following integer data:
df1, df2, df3, df4, df5 = (pd.DataFrame(rng.integers(0, 1000, (100, 3)))
for i in range(5))
Arithmetic operators#
pd.eval supports all arithmetic operators. For example:
result1 = -df1 * df2 / (df3 + df4) - df5
result2 = pd.eval('-df1 * df2 / (df3 + df4) - df5')
np.allclose(result1, result2)
True
Comparison operators#
pd.eval supports all comparison operators, including chained expressions:
result1 = (df1 < df2) & (df2 <= df3) & (df3 != df4)
result2 = pd.eval('df1 < df2 <= df3 != df4')
np.allclose(result1, result2)
True
Bitwise operators#
pd.eval supports the & and | bitwise operators:
result1 = (df1 < 0.5) & (df2 < 0.5) | (df3 < df4)
result2 = pd.eval('(df1 < 0.5) & (df2 < 0.5) | (df3 < df4)')
np.allclose(result1, result2)
True
In addition, it supports the use of the literal and and or in Boolean expressions:
result3 = pd.eval('(df1 < 0.5) and (df2 < 0.5) or (df3 < df4)')
np.allclose(result1, result3)
True
Object attributes and indices#
pd.eval supports access to object attributes via the obj.attr syntax and indexes via the obj[index] syntax:
result1 = df2.T[0] + df3.iloc[1]
result2 = pd.eval('df2.T[0] + df3.iloc[1]')
np.allclose(result1, result2)
True
Other operations#
Other operations, such as function calls, conditional statements, loops, and other more involved constructs are currently not implemented in pd.eval.
If you’d like to execute these more complicated types of expressions, you can use the NumExpr library itself.
DataFrame.eval for Column-Wise Operations#
Just as Pandas has a top-level pd.eval function, DataFrame objects have an eval method that works in similar ways.
The benefit of the eval method is that columns can be referred to by name.
We’ll use this labeled array as an example:
df = pd.DataFrame(rng.random((1000, 3)), columns=['A', 'B', 'C'])
df.head()
| A | B | C | |
|---|---|---|---|
| 0 | 0.850888 | 0.966709 | 0.958690 |
| 1 | 0.820126 | 0.385686 | 0.061402 |
| 2 | 0.059729 | 0.831768 | 0.652259 |
| 3 | 0.244774 | 0.140322 | 0.041711 |
| 4 | 0.818205 | 0.753384 | 0.578851 |
Using pd.eval as in the previous section, we can compute expressions with the three columns like this:
result1 = (df['A'] + df['B']) / (df['C'] - 1)
result2 = pd.eval("(df.A + df.B) / (df.C - 1)")
np.allclose(result1, result2)
True
The DataFrame.eval method allows much more succinct evaluation of expressions with the columns:
result3 = df.eval('(A + B) / (C - 1)')
np.allclose(result1, result3)
True
Notice here that we treat column names as variables within the evaluated expression, and the result is what we would wish.
Assignment in DataFrame.eval#
In addition to the options just discussed, DataFrame.eval also allows assignment to any column.
Let’s use the DataFrame from before, which has columns 'A', 'B', and 'C':
df.head()
| A | B | C | |
|---|---|---|---|
| 0 | 0.850888 | 0.966709 | 0.958690 |
| 1 | 0.820126 | 0.385686 | 0.061402 |
| 2 | 0.059729 | 0.831768 | 0.652259 |
| 3 | 0.244774 | 0.140322 | 0.041711 |
| 4 | 0.818205 | 0.753384 | 0.578851 |
We can use df.eval to create a new column 'D' and assign to it a value computed from the other columns:
df.eval('D = (A + B) / C', inplace=True)
df.head()
| A | B | C | D | |
|---|---|---|---|---|
| 0 | 0.850888 | 0.966709 | 0.958690 | 1.895916 |
| 1 | 0.820126 | 0.385686 | 0.061402 | 19.638139 |
| 2 | 0.059729 | 0.831768 | 0.652259 | 1.366782 |
| 3 | 0.244774 | 0.140322 | 0.041711 | 9.232370 |
| 4 | 0.818205 | 0.753384 | 0.578851 | 2.715013 |
In the same way, any existing column can be modified:
df.eval('D = (A - B) / C', inplace=True)
df.head()
| A | B | C | D | |
|---|---|---|---|---|
| 0 | 0.850888 | 0.966709 | 0.958690 | -0.120812 |
| 1 | 0.820126 | 0.385686 | 0.061402 | 7.075399 |
| 2 | 0.059729 | 0.831768 | 0.652259 | -1.183638 |
| 3 | 0.244774 | 0.140322 | 0.041711 | 2.504142 |
| 4 | 0.818205 | 0.753384 | 0.578851 | 0.111982 |
Local Variables in DataFrame.eval#
The DataFrame.eval method supports an additional syntax that lets it work with local Python variables.
Consider the following:
column_mean = df.mean(1)
result1 = df['A'] + column_mean
result2 = df.eval('A + @column_mean')
np.allclose(result1, result2)
True
The @ character here marks a variable name rather than a column name, and lets you efficiently evaluate expressions involving the two “namespaces”: the namespace of columns, and the namespace of Python objects.
Notice that this @ character is only supported by the DataFrame.eval method, not by the pandas.eval function, because the pandas.eval function only has access to the one (Python) namespace.
The DataFrame.query Method#
The DataFrame has another method based on evaluated strings, called query.
Consider the following:
result1 = df[(df.A < 0.5) & (df.B < 0.5)]
result2 = pd.eval('df[(df.A < 0.5) & (df.B < 0.5)]')
np.allclose(result1, result2)
True
As with the example used in our discussion of DataFrame.eval, this is an expression involving columns of the DataFrame.
However, it cannot be expressed using the DataFrame.eval syntax!
Instead, for this type of filtering operation, you can use the query method:
result2 = df.query('A < 0.5 and B < 0.5')
np.allclose(result1, result2)
True
In addition to being a more efficient computation, compared to the masking expression this is much easier to read and understand.
Note that the query method also accepts the @ flag to mark local variables:
Cmean = df['C'].mean()
result1 = df[(df.A < Cmean) & (df.B < Cmean)]
result2 = df.query('A < @Cmean and B < @Cmean')
np.allclose(result1, result2)
True
Performance: When to Use These Functions#
When considering whether to use eval and query, there are two considerations: computation time and memory use.
Memory use is the most predictable aspect. As already mentioned, every compound expression involving NumPy arrays or Pandas DataFrames will result in implicit creation of temporary arrays. For example, this:
x = df[(df.A < 0.5) & (df.B < 0.5)]
is roughly equivalent to this:
tmp1 = df.A < 0.5
tmp2 = df.B < 0.5
tmp3 = tmp1 & tmp2
x = df[tmp3]
If the size of the temporary DataFrames is significant compared to your available system memory (typically several gigabytes), then it’s a good idea to use an eval or query expression.
You can check the approximate size of your array in bytes using this:
df.values.nbytes
32000
On the performance side, eval can be faster even when you are not maxing out your system memory.
The issue is how your temporary objects compare to the size of the L1 or L2 CPU cache on your system (typically a few megabytes); if they are much bigger, then eval can avoid some potentially slow movement of values between the different memory caches.
In practice, I find that the difference in computation time between the traditional methods and the eval/query method is usually not significant—if anything, the traditional method is faster for smaller arrays!
The benefit of eval/query is mainly in the saved memory, and the sometimes cleaner syntax they offer.
We’ve covered most of the details of eval and query here; for more information on these, you can refer to the Pandas documentation.
In particular, different parsers and engines can be specified for running these queries; for details on this, see the discussion within the “Enhancing Performance” section of the documentation.