2013 survey

lib/salary_averager.rb

@@ -1,46 +1,138 @@
 require 'rubygems'
-require 'faster_csv'
+require 'csv'
 
+# Usage:
+# file_path = File.expand_path('data/data2013.csv', File.dirname(__FILE__))
+# puts SalaryAverager.new(file_path).salary_report
 class SalaryAverager
 
-  def initialize(path_to_data)
-    @data = FasterCSV.read(path_to_data)
-    @data.shift
+  class Averager < Array
+    def average
+      return nil if empty?
+      inject(&:+) / size
+    end
+
+    def median
+      return nil if empty?
+      if size.odd?
+        sort[size / 2]
+      else
+        (sort[size / 2 - 1] + sort[size / 2]) / 2
+      end
+    end
+
+    def report
+      return too_small_message if sample_size_too_small?
+      "#{size} respondents.  Average: #{average}.   Median: #{median}."
+    end
+
+    def sample_size_too_small?
+      size < 3
+    end
+
+    def too_small_message
+      case count = size
+      when 0
+        "No respondents!"
+      when 1, 2
+        "Sample size too small for anonymous reporting."
+      end
+    end
+  end
+
+
+  class MalformedSpreadsheetError < ArgumentError
+    def message
+      msg = "Spreadsheet must have column headers that include 'work status' or 'freelance',"
+      msg += "'manage' or 'tell other developers', and 'salary'.\n"
+      msg + super()
+    end
+  end
+
+
+  def initialize(file_path)
+    @data = CSV.read(file_path)
+    process_headers(@data.shift)
+  end
+
+  def process_headers(headers)
+    @employment_column = headers.find_index { |entry| /(work status|freelance)/i =~ entry }
+    @manager_column    = headers.find_index { |entry| /(manage|tell other developers)/i =~ entry }
+    @skill_column      = headers.find_index { |entry| /(skill|ability)/i =~ entry }
+    @salary_column     = headers.find_index { |entry| /salary/i =~ entry }
+
+    unless @employment_column && @manager_column && @salary_column # skill column is optional
+      e = MalformedSpreadsheetError.new
+      raise e
+    end
+  end
+
+  def entry_count
+    @data.size
+  end
+
+  def freelancers
+    @data.select { |row| /(freelance|Yes)/i =~ row[@employment_column] }
+  end
+
+  def non_freelancers
+    @data.reject { |row| /(freelance|Yes)/i =~ row[@employment_column] }
+  end
+
+  def managers(data = @data)
+    data.select { |row| row[@manager_column] == "Yes" }
+  end
+
+  def non_managers(data = @data)
+    data.select { |row| row[@manager_column] == "No" }
+  end
+
+  def salaries_from(rows)
+    rows.map { |row| row[@salary_column].to_i }
+  end
+
+  def skill_ratings
+    if @skill_column
+      @data.map { |row| row[@skill_column] }.map(&:to_f)
+    end
   end
 
   def freelancer_salaries
-    @data.select { |entry| entry[4] == "Yes" }.map { |entry| entry[1] }.map(&:to_f)
+    salaries_from(freelancers)
   end
 
   def non_freelancer_salaries
-    @data.select { |entry| entry[4] == "No" }.map { |entry| entry[1] }.map(&:to_f)
+    salaries_from(non_freelancers)
   end
 
   def manager_salaries
-    @data.select { |entry| entry[3] == "Yes" }.map { |entry| entry[1] }.map(&:to_f)
+    salaries_from(managers)
   end
 
   def non_manager_salaries
-    @data.select { |entry| entry[3] == "No" }.map { |entry| entry[1] }.map(&:to_f)
+    salaries_from(non_managers)
   end
 
-  def skill_ratings
-    @data.map { |entry| entry[2] }.map(&:to_f)
+  def non_manager_non_freelancer_salaries
+    salaries_from(non_managers(non_freelancers))
   end
 
-end
-
-class Array
-  def average
-    self.inject(&:+) / self.size.to_f
+  def manager_non_freelancer_salaries
+    salaries_from(managers(non_freelancers))
   end
-end
 
+  def report(salaries)
+    Averager.new(salaries).report
+  end
 
-sa = SalaryAverager.new('data/data.csv')
+  def salary_report
+    "#{entry_count} people responded to the survey.\n" <<
+      "Freelancers:     #{report(freelancer_salaries)}\n" <<
+      "Non-freelancers: #{report(non_freelancer_salaries)}\n" <<
+      "Managers:        #{report(manager_salaries)}\n" <<
+      "Non-managers:    #{report(non_manager_salaries)}\n" <<
+      "Manager non-freelancers:        #{report(manager_non_freelancer_salaries)}\n" <<
+      "Non-manager non-freelancers:    #{report(non_manager_non_freelancer_salaries)}"
+  end
 
-puts "Average freelancer salary: %d"     % sa.freelancer_salaries.average
-puts "Average non-freelancer salary: %d" % sa.non_freelancer_salaries.average
-puts "Average manager salary: %d"        % sa.manager_salaries.average
-puts "Average non-manager salary: %d"    % sa.non_manager_salaries.average
-puts "Average skill rating: %d"    % sa.skill_ratings.average
+end

spec/fake_data2013.csv

@@ -0,0 +1,5 @@
+"Entry Id","What is your current annual salary? (Please estimate and include any annual bonuses).","Please rate your Rails ability from 1 (complete newbie) to 10 (Core Team Member).","Do you tell other developers what to do? (Are you a project manager or team lead?)","Are you a freelancer?","Any comments you'd like to make?","Date Created","Created By","Last Updated","Updated By","IP Address","Last Page Accessed","Completion Status"
+"1","100000","6","Yes","Full-time, salaried employee","","2011-03-05 13:38:50","public","","","74.104.152.233","1","1"
+"2","110000","6","No","Freelancer","","2011-03-05 13:38:50","public","","","74.104.152.233","1","1"
+"3","120000","7","Yes","freelancer","","2011-03-05 13:39:50","public","","","74.104.152.234","1","1"
+"4","130000","8","Yes","full-time, hourly","","2011-03-05 13:39:50","public","","","74.104.152.234","1","1"

spec/numbers_only.csv

@@ -0,0 +1,4 @@
+"1","100000","6","Yes","Full-time, salaried employee","","2011-03-05 13:38:50","public","","","74.104.152.233","1","1"
+"2","110000","6","No","Freelancer","","2011-03-05 13:38:50","public","","","74.104.152.233","1","1"
+"3","120000","7","Yes","freelancer","","2011-03-05 13:39:50","public","","","74.104.152.234","1","1"
+"4","130000","8","Yes","full-time, hourly","","2011-03-05 13:39:50","public","","","74.104.152.234","1","1"

spec/salary_averager_spec.rb

@@ -3,45 +3,81 @@
 require 'salary_averager'
 
 describe SalaryAverager do
-  before(:each) do
-    @averager = SalaryAverager.new("spec/fake_data.csv") 
-  end
+  let(:averager) { SalaryAverager.new("spec/fake_data2013.csv") }
 
   describe "#freelancer_salaries" do
     it "returns all salaries for people who are freelancers" do
-      @averager.freelancer_salaries.should == [110000.0, 120000.0]
+      expect(averager.freelancer_salaries).to eq([110000, 120000])
     end
   end
 
   describe "#non_freelancer_salaries" do
     it "returns all salaries for people who are not freelancers" do
-      @averager.non_freelancer_salaries.should == [100000.0, 130000.0]
+      expect(averager.non_freelancer_salaries).to eq([100000, 130000])
     end
   end
 
   describe "#manager_salaries" do
     it "returns all salaries for people who manage other devs" do
-      @averager.manager_salaries.should == [100000.0, 120000.0, 130000.0]
+      expect(averager.manager_salaries).to eq([100000, 120000, 130000])
     end
   end
 
   describe "#non_manager_salaries" do
     it "returns all salaries for people who do not manage other devs" do
-      @averager.non_manager_salaries.should == [110000.0]
+      expect(averager.non_manager_salaries).to eq([110000])
     end
   end
 
-  describe "#skill_ratings" do 
+  describe "#skill_ratings" do
     it "returns the skill ratings for everyone" do
-      @averager.skill_ratings.should == [6,6,7,8]
+      expect(averager.skill_ratings).to eq([6,6,7,8])
     end
   end
 
+  it "errors on a spreadsheet without the right headers" do
+    expect {
+      SalaryAverager.new("spec/numbers_only.csv")
+    }.to raise_error
+  end
+
 end
 
-describe "Goofy Array monkeypatch" do
-  it "averages the contents of an array" do
-    [2,4,6].average.should == 4
+describe "SalaryAverager::Averager" do
+  it "averages correctly" do
+    expect(SalaryAverager::Averager.new([2,4,12]).average).to eq(6)
+  end
+
+  it "calculates median correctly on odd-length arrays" do
+    expect(SalaryAverager::Averager.new([2,4,12]).median).to eq(4)
+  end
+
+  it "calculates median correctly on even-length arrays" do
+    expect(SalaryAverager::Averager.new([2,4,12,48]).median).to eq(8)
   end
+
+  it "averages for two, one, or no people" do
+    expect(SalaryAverager::Averager.new([2,12]).average).to eq(7)
+    expect(SalaryAverager::Averager.new([2]).average).to eq(2)
+    expect { SalaryAverager::Averager.new([]).average }.not_to raise_exception
+    expect(SalaryAverager::Averager.new([]).average).to be_nil
+  end
+
+  it "calculates median for two, one, or no people" do
+    expect(SalaryAverager::Averager.new([2,12]).median).to eq(7)
+    expect(SalaryAverager::Averager.new([2]).median).to eq(2)
+    expect {
+      SalaryAverager::Averager.new([]).median
+    }.not_to raise_exception
+    expect(SalaryAverager::Averager.new([]).median).to be_nil
+  end
+
+
+  it "doesn't report on small arrays" do
+    expect(SalaryAverager::Averager.new([2,4]).report).to match(/too small/)
+    expect(SalaryAverager::Averager.new([2]).report).to match(/too small/)
+    expect(SalaryAverager::Averager.new([]).report).to match(/(no |none)/i)
+  end
+
 end