Statistics

This ActiverRecord plugin allows you to easily define and pull statistics for AR models. This plugin was built with reporting in mind.

!!! READ ANNOUCEMENT BY TAM !!!

I need this specific behaviour for collect staticitics in time ranges on specific field (created_at, payed_at etc..). If you decide to use my modification:

gem 'statistics', github: "tam-vo/statistics"

You will be no able to use named filters:

[ :range_today :range_week, :range_month and :range_year ]

If you want to use it here is how:

class User < ActiveRecord::Base
  define_statistic :user_count, :count => :all
end

# I want count of user registrated last week and month
last_week = User.get_stat(:user_count, :range_week => :created_at)
last_month = User.get_stat(:user_count, :range_month => :created_at)

Allways use: :range_week => :created_at

Announcements

• Bug: There is a bug in Rails 2.x where grouping by multiple fields results in wrong values for calculations. I have created a patch for this bug, please encourage the fix by giving feedback or giving +1 plus a short supportive comment. The patch lives here: https://rails.lighthouseapp.com/projects/8994/tickets/5182-activerecordcalculations-returns-incorrect-data-when-grouping-by-multiple-fields.

Installation

gem install statistics

OR script/plugin install git://github.com/tam-vo/statistics.git

Examples

Defining statistics is similar to defining named scopes. Strings and symbols both work as names.

class Account < ActiveRecord::Base
  define_statistic :user_count, :count => :all
  define_statistic :average_age, :average => :all, :column_name => 'age'
  define_statistic 'subscriber count', :count => :all, :conditions => "subscription_opt_in = 1"
end

class Donations < ActiveRecord::Base
  define_statistic :total_donations, :sum => :all, :column_name => "amount"
end

Actually pulling the numbers is simple:

#####for all stats

Account.statistics                   # returns { :user_count => 120, :average_age => 28, 'subscriber count' => 74 }

#####for a single stat

Account.get_stat(:user_count)      # returns 120

Here are some additional benefits of using this plugin:

Easily Filter

Note: I found filtering to be an important part of reporting (ie. filtering by date). All filters are optional so even if you define them you don’t have to use them when pulling data. Using the filter_all_stats_on method and :joins options you can make things filterable by the same things which I found to be extremely useful.

class Account < ActiveRecord::Base
  define_statistic :user_count, :count => :all, :filter_on => { :state => 'state = ?', :created_after => 'DATE(created_at) > ?'}
  define_statistic :subscriber_count, :count => :all, :conditions => "subscription_opt_in = true"

  filter_all_stats_on(:account_type, "account_type = ?")
end

Account.statistics(:account_type => 'non-admin')
Account.get_stat(:user_count,  :account_type => 'non-admin',  :created_after => ‘2009-01-01’, :state => 'NY')

# NOTE: filters are optional (ie. no filters will be applied if none are passed in)
Account.get_stat(:user_count)

Caching

This is a new feature that uses Rails.cache. You can cache certain statistics for a specified amount of time (see below). By default caching is disabled if you do not pass in the :cache_for option. It is also important to note that caching is scoped by filters, there is no way around this since different filters produce different values. class Account < ActiveRecord::Base define_statistic :user_count, :count => :all, :cache_for => 30.minutes, :filter_on { :state => 'state = ?' } end

Account.statistics(:state => 'NY') # This call generates a SQL query

Account.statistics(:state => 'NY') # This call and subsequent calls for the next 30 minutes will use the cached value

Account.statistics(:state => 'PA') # This call generates a SQL query because the user count for NY and PA could be different (and probably is)

Note: If you want Rails.cache to work properly, you need to use mem_cache_store in your rails enviroment file (ie. config.cache_store = :mem_cache_store in your enviroment.rb file).

Standardized

All ActiveRecord classes now respond to statistics and get_stat methods

all_stats = []
[ Account, Post, Comment ].each do |ar|
  all_stats << ar.statistics
end

Calculated statistics (DRY)

You can define calculated metrics in order to perform mathematical calculations on one or more defined statistics.

class Account < ActiveRecord::Base
  has_many :donations

  define_statistic :user_count, :count => :all
  define_statistic :total_donations, :sum => :all, :column_name => 'donations.amount', :joins => :donations

  define_calculated_statistic :average_donation_per_user do
    defined_stats(:total_donations) / defined_stats(:user_count)
  end

  filter_all_stats_on(:account_type, "account_type = ?")
  filter_all_stats_on(:state, "state = ?")
  filter_all_stats_on(:created_after, "DATE(created_at) > ?")
end

Pulling stats for calculated metrics is the same as for regular statistics. They also work with filters like regular statistics!

Account.get_stat(:average_donation_per_user, :account_type => 'non-admin', :state => 'NY')
Account.get_stat(:average_donation_per_user, :created_after => '2009-01-01')

Reuse scopes you already have defined

You can reuse the code you have written to do reporting.

class Account < ActiveRecord::Base
  has_many :posts

  named_scope :not_admins, :conditions => “account_type = ‘non-admin’”
  named_scope :accounts_with_posts, :joins => :posts

  define_statistic :active_users_count, :count => [:not_admins, :accounts_with_posts]
end

Accepts all ActiveRecord::Calculations options

The :conditions and :joins options are all particularly useful

class Account < ActiveRecord::Base
  has_many :posts

  define_statistic :active_users_count, :count => :all, :joins => :posts, :conditions => "account_type = 'non-admin'"
end

Copyright (c) 2009 Alexandru Catighera, released under MIT license