Python Data Analysis
from Start to Finish

Move Fast


siuba (小巴) is an analysis framework that makes data science faster. It provides a simple, consistent interface that handles messy, real-life data.


Get Started

Write Once, Run Everywhere


Use the same code to work directly with a pandas DataFrame, or execute SQL queries against a database—such as postgresql, duckdb or snowflake.


See Backends

Learn Together


Everybody has to start somewhere. Bring your questions, bring your hot takes, bring the excel spreadsheets your boss emailed you.


Join Community

Key Features

Five core table functions

from siuba.data import cars
from siuba import _, filter

filter(cars, _.mpg == _.mpg.max())
cyl mpg hp
19 4 33.9 65 
from siuba.data import cars
from siuba import _, arrange

arrange(cars, _.mpg)
cyl mpg hp
14 8 10.4 205
15 8 10.4 215
... ... ... ...
17 4 32.4 66
19 4 33.9 65

32 rows × 3 columns

from siuba.data import cars
from siuba import _, select

select(cars, ~_.hp)
cyl mpg
0 6 21.0
1 6 21.0
... ... ...
30 8 15.0
31 4 21.4

32 rows × 2 columns

from siuba.data import cars
from siuba import _, mutate

mutate(cars, demeaned = _.hp - _.hp.mean())
cyl mpg hp demeaned
0 6 21.0 110 -36.6875
1 6 21.0 110 -36.6875
... ... ... ... ...
30 8 15.0 335 188.3125
31 4 21.4 109 -37.6875

32 rows × 4 columns

from siuba.data import cars
from siuba import _, group_by, summarize

summarize(cars, hp_per_cyl = (_.hp / _.cyl).mean())
hp_per_cyl
0 23.001302

Everything works with group_by()

The pipe (>>) puts it all together

from siuba.data import mtcars
from siuba import _, group_by, filter

(mtcars 
    >> group_by(_.cyl)
    >> filter(_.hp == _.hp.max())
)

(grouped data frame)

mpg cyl disp hp drat wt qsec vs am gear carb
27 30.4 4 95.1 113 3.77 1.513 16.9 1 1 5 2
29 19.7 6 145.0 175 3.62 2.770 15.5 0 1 5 6
30 15.0 8 301.0 335 3.54 3.570 14.6 0 1 5 8

Get Started Examples

Write Once, Run Everywhere

from siuba import _, tbl, group_by, filter
from siuba.data import cars

# setup ----





# analysis ----
(cars
    >> group_by(_.cyl)
    >> filter(_.mpg < _.mpg.mean())
)

(grouped data frame)

cyl mpg hp
2 4 22.8 93
5 6 18.1 105
... ... ... ...
30 8 15.0 335
31 4 21.4 109

16 rows × 3 columns

from siuba import _, tbl, group_by, filter
from siuba.data import cars

# setup ----
from sqlalchemy import create_engine

engine = create_engine("duckdb:///:memory:")


# analysis ----
(tbl(engine, "cars", cars)
    >> group_by(_.cyl)
    >> filter(_.mpg < _.mpg.mean())
)
# Source: lazy query
# DB Conn: Engine(duckdb:///:memory:)
# Preview:
cyl mpg hp
0 4 22.8 93
1 4 24.4 62
2 4 22.8 95
3 4 21.5 97
4 4 26.0 91

# .. may have more rows

from siuba import _, tbl, group_by, filter
from siuba.data import cars

# setup ----
from sqlalchemy import create_engine

engine = create_engine("sqlite:///:memory:")
cars.to_sql("cars", engine, index=False)

# analysis ----
(tbl(engine, "cars")
    >> group_by(_.cyl)
    >> filter(_.mpg < _.mpg.mean())
)
# Source: lazy query
# DB Conn: Engine(sqlite:///:memory:)
# Preview:
cyl mpg hp
0 4 22.8 93
1 4 24.4 62
2 4 22.8 95
3 4 21.5 97
4 4 26.0 91

# .. may have more rows

SQL Backend Quickstart Support Table

Examples

See the examples page.

Community

Visit siuba on github, or open an issue with a bug or feature request.