This package provides utilities for turning Django Querysets into Pivot-Tables and Histograms by letting your database do all the heavy lifting.
I am going to shamelessly lift examples from the wikipedia page referenced in the header. Here is part of the table of shirt sales:
Region | Gender | Style | Ship Date | Units | Price | Cost |
---|---|---|---|---|---|---|
East | Boy | Tee | 1/31/2005 | 12 | 11.04 | 10.42 |
East | Boy | Golf | 1/31/2005 | 12 | 13 | 12.6 |
East | Boy | Fancy | 1/31/2005 | 12 | 11.96 | 11.74 |
East | Girl | Tee | 1/31/2005 | 10 | 11.27 | 10.56 |
East | Girl | Golf | 1/31/2005 | 10 | 12.12 | 11.95 |
East | Girl | Fancy | 1/31/2005 | 10 | 13.74 | 13.33 |
West | Boy | Tee | 1/31/2005 | 11 | 11.44 | 10.94 |
West | Boy | Golf | 1/31/2005 | 11 | 12.63 | 11.73 |
West | Boy | Fancy | 1/31/2005 | 11 | 12.06 | 11.51 |
West | Girl | Tee | 1/31/2005 | 15 | 13.42 | 13.29 |
West | Girl | Golf | 1/31/2005 | 15 | 11.48 | 10.67 |
Etc. |
We might want to know How many Units did we sell in each Region for every Ship Date? And get a result like:
Region | 1/31/2005 | 2/1/2005 | 2/2/2005 | 2/3/2005 | 2/4/2005 |
---|---|---|---|---|---|
East | 66 | 80 | 102 | 93 | 114 |
North | 86 | 91 | 95 | 88 | 107 |
South | 73 | 78 | 84 | 76 | 91 |
West | 92 | 103 | 111 | 104 | 123 |
It takes 3 quantities to pivot the original table into the summary result, two columns and an aggregate of a third column. In this case the two columns are Region and Ship Date, the third column is Units and the aggregate is Sum
The pivot function
Pivot tables are generated by the pivot function, which takes a Model and 3 attribute names, to make a pivot table like the example above:
>>> pivot_table = pivot(ShirtSales, 'shipped', 'region', 'units')
The result is a ValuesQuerySet, which means the objects returned are dictionaries. Each dictionary has a key for the row ('shipped' dates in this case) and a key for every value of the column ('region' in this case).
>>> for record in pivot_table:
... print record
... {u'West': 59, 'shipped': datetime.date(2004, 12, 24), u'East': 71, u'North': 115, u'South': 56}
... {u'West': 55, 'shipped': datetime.date(2005, 1, 31), u'East': 65, u'North': 121, u'South': 66}
... {u'West': 56, 'shipped': datetime.date(2005, 2, 1), u'East': 62, u'North': 124, u'South': 68}
... {u'West': 56, 'shipped': datetime.date(2005, 2, 2), u'East': 59, u'North': 127, u'South': 71}
... {u'West': 66, 'shipped': datetime.date(2005, 3, 1), u'East': 55, u'North': 131, u'South': 65}
... {u'West': 68, 'shipped': datetime.date(2005, 3, 2), u'East': 56, u'North': 130, u'South': 62}
... {u'West': 71, 'shipped': datetime.date(2005, 4, 3), u'East': 56, u'North': 130, u'South': 59}
... {u'West': 65, 'shipped': datetime.date(2005, 5, 6), u'East': 66, u'North': 120, u'South': 55}
The first argument can be a Model, QuerySet, or Manager. This allows you to generate a pivot table filtered by another column. For example, you may want to know how many units were sold in each region for every shipped date, but only for Golf shirts:
>>> pivot_table = pivot(ShirtSales.objects.filter(style='Golf'), 'region', 'shipped', 'units')
The pivot function takes an option parameter for how to aggregate the data. For example, instead of the total units sold in each region for every ship date, we might be interested in the average number of units per order. Then we can pass the Avg aggregation function
>>> from django.db.models import Avg
>>> pivot_table = pivot(ShirtSales, 'region', 'shipped', 'units', aggregation=Avg)
If your data is stored across multiple tables, just use Django's double underscore notation to traverse foreign key relationships. For example, instead of the ShirtSales model having a region attribute, it might have a foreign key to a Store model, which in turn has a foreign key to a Region model, which has an attribute called name. Then our pivot call looks like
>>> pivot_table = pivot(ShirtSales, 'store__region__name', 'shipped', 'units')
It's also possible that the data column we are aggregating over should be a computed column. In our example ShirtSales model we are storing the number of units and the price per unit, but not the total cost of the order. If we want to know the average order size in dollars in each region for every ship date, we can pivot the ShirtSales table:
>>> from django.db.models import F, Avg
>>> pivot_table = pivot(ShirtSales, 'region', 'shipped', F('units') * F('price'), Avg)
The histogram function
This library also supports creating histograms from a single column of data with the histogram function, which takes a Model, a single attribute name and an iterable of left edges of bins.
>>> hist = histogram(ShirtSales, 'units', bins=[0, 10, 15])
Like pivot, the first argument can be a Model, QuerySet, or Manager. The result is a list of dictionaries:
>>> hist
[{'bin': '0', 'units': 0},
{'bin': '10', 'units': 0},
{'bin': '15', 'units': 0}]
It's also possible to get several histograms from a single query by slicing the data on one
of the columns. For example, instead of the histogram above, we might want two histograms,
one for boys and one for girls. The gender
column of ShirtSales
has two values,
'Boy'
and 'Girl'
. Passing the gender column as a 4th optional parameter to histogram
will slice the data on that column.
>>> hist = histogram(ShirtSales, 'units', bins=[0, 10, 15], slice_on='gender')
The result is a ValuesQuerySet where each row corresponds to one bin
>>> for row in hist:
print row
{'bin': u'0', u'Boy': 53, u'Girl': 53}
{'bin': u'10', u'Boy': 40, u'Girl': 41}
{'bin': u'15', u'Boy': 27, u'Girl': 26}
Just:
pip install django-pivot
put django_pivot in installed apps in your settings file, and then you:
from django_pivot.pivot import pivot from django_pivot.histogram import histogram
And off you go.
The test suite is run by Travis with Django versions 1.10 and 1.11 and backends sqlite, MySQL, and Postgres. If you want to run the test suite locally, from the root directory:
python runtests.py --settings=django_pivot.tests.test_sqlite_settings
That will use sqlite as the backend and whatever version of Django you have in your current environment.
MIT
Copyright 2017 Brad Martsberger
Permission is hereby granted, free of charge, to any person obtaining a copy of this software and associated documentation files (the "Software"), to deal in the Software without restriction, including without limitation the rights to use, copy, modify, merge, publish, distribute, sublicense, and/or sell copies of the Software, and to permit persons to whom the Software is furnished to do so, subject to the following conditions:
The above copyright notice and this permission notice shall be included in all copies or substantial portions of the Software.
THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.