Statistics storage for Django
Keep track of your statistics.
- Source code
The following concepts are used:
A piece of information to keep track of. For example, “Order count”, or “Number of users signed up”.
Metrics are organized in groups, each group is called a domain. For example you can have a “shopping” domain with metrics such as “Order count”, “Items sold”, “Products viewed”, and a “users” domain with “Login count”, “Signup count”. Or, in case you are tracking external statistics from social networks, you may introduce a “Twitter” domain, and metrics “Followers count”.
Used to store the actual values by date, for a specific metric.
The time period for which the stored value holds. For example, you can keep track of cumulative, all-time, numbers (Period.LIFETIME), store incremental values on a daily basis (Period.DAY), or keep track of a rolling count for the last 7 days (Period.WEEK).
- Reference IDs
Domains and metrics must be assigned unique reference IDs (of type string). Rationale: Having a human readable, non PK based, reference is esential as soon as you are going to export statistics.
First, setup your domains:
from trackstats.models import Domain Domain.objects.SHOPPING = Domain.objects.register( ref='shopping', name='Shopping') Domain.objects.USERS = Domain.object.register( ref='users', name='Users') Domain.objects.TWITTER = Domain.object.register( ref='twitter', name='Twitter')
Define a few metrics:
from trackstats.models import Domain, Metric Metric.objects.SHOPPING_ORDER_COUNT = Metric.objects.register( domain=Domain.objects.SHOPPING, ref='order_count', name='Number of orders sold') Metric.objects.USERS_USER_COUNT = Metric.objects.register( domain=Domain.objects.USERS, ref='user_count', name='Number of users signed up') Metric.objects.TWITTER_FOLLOWER = Metric.objects.register( # Matches Twitter API ref='followers_count', domain=Domain.objects.TWITTER)
Now, let’s store some one-off statistics:
from trackstats.models import StatisticByDate, Domain, Metric, Period # All-time, cumulative, statistic n = Order.objects.all().count() StatisticByDate.objects.record( metric=Metric.objects.SHOPPING_ORDER_COUNT, value=n, Period=Period.LIFETIME) # Users signed up, at a specific date dt = date.today() n = User.objects.filter( date_joined__day=dt.day, date_joined__month=dt.month, date_joined__year=dt.year).count() StatisticByDate.objects.record( metric=Metric.objects.USERS_USER_COUNT, value=n, Period=Period.DAY)
Creating code to store statistics yourself can be a tedious job. Luckily, a few shortcuts are available to track statistics without having to write any code yourself.
Consider you want to keep track of the number of comments created on a daily basis:
from trackstats.trackers import CountObjectsByDateTracker CountObjectsByDateTracker( period=Period.DAY, metric=Metric.objects.COMMENT_COUNT, date_field='timestamp').track(Comment.objects.all())
Or, in case you want to track the number of comments, per user, on a daily basis:
CountObjectsByDateAndObjectTracker( period=Period.DAY, metric=Metric.objects.COMMENT_COUNT, # comment.user points to a User object_model=User, object_field='user', # Comment.timestamp is used for grouping date_field='timestamp').track(Comment.objects.all())
The StatisticByDate model represents statistics grouped by date – the most common use case.
Another common use case is to group by both date and some other object (e.g. a user, category, site). For this, use StatisticByDateAndObject. It uses a generic foreign key.
If you need to group in a different manner, e.g. by country, province and date, you can use the AbstractStatistic base class to build just that.
If you like this, you may also like:
Download the file for your platform. If you're not sure which to choose, learn more about installing packages.