benchmark written for Daru::Vector#uniq and comparion with Pandas and…

… NumPy

Indentation to 2 spaces

benchmarks/statistics/benchmark.rb

@@ -16,10 +16,12 @@
 	Benchmarker.benchmark_median_absolute_deviation()
 	Benchmarker.benchmark_sum_of_squared_deviation()
 	Benchmarker.benchmark_average_deviation_population()
+	Benchmarker.benchmark_unique()
 	puts 
 end
 
-Benchmarker.result_with_size()
+# Benchmarker.result_with_size() Uncomment it to see only Daru::Vector benchmark.
+Benchmarker.result_compare()
 
 # Real times for vector size [10**2, 10**3,10**4,10**5,10**6] 
 

benchmarks/statistics/benchmarker.rb

@@ -17,6 +17,7 @@ class Benchmarker
 	@result_sum_of_squared_deviation = []
 	@result_average_deviation_populationa = []
 	@result_create_df = []
+	@result_unique = []
 
 	class << self
 		attr_accessor :df, :df_size, :result_mean,	:result_mode,
@@ -26,14 +27,15 @@ class << self
 		:result_median_absolute_deviation,
 		:result_sum_of_squared_deviation,
 		:result_average_deviation_populationa,
-		:result_create_df
+		:result_create_df,
+		:result_unique
 	end
 
 	def self.benchmark_mean()
 
 		bench = Benchmark.bm do |x|
 			report = x.report('Vector mean => ') do
-		   		@df[0].mean
+				@df[0].mean
 			end
 			@result_mean.append("%1.20f" % report.real)
 			# print "%1.20f" % report.real
@@ -44,7 +46,7 @@ def self.benchmark_mode()
 
 		bench = Benchmark.bm do |x|
 			report =x.report('Vector mode => ') do
-		   		@df[0].mode
+				@df[0].mode
 			end
 			@result_mode.append("%1.20f" % report.real)
 		end	
@@ -54,7 +56,7 @@ def self.benchmark_median()
 
 		bench = Benchmark.bm do |x|
 			report =x.report('Vector median => ') do
-		   		@df[0].median
+				@df[0].median
 			end
 			@result_median.append("%1.20f" % report.real)
 		end	
@@ -64,7 +66,7 @@ def self.benchmark_sum()
 
 		bench = Benchmark.bm do |x|
 			report = x.report('Vector sum => ') do
-		   		@df[0].sum
+				@df[0].sum
 			end
 			@result_sum.append("%1.20f" % report.real)
 		end	
@@ -74,7 +76,7 @@ def self.benchmark_product()
 
 		bench = Benchmark.bm do |x|
 			report = x.report('Vector product => ') do
-		   		@df[0].product
+				@df[0].product
 			end
 			@result_product.append("%1.20f" % report.real)
 		end	
@@ -84,7 +86,7 @@ def self.benchmark_median_absolute_deviation()
 
 		bench = Benchmark.bm do |x|
 			report = x.report('Vector median_absolute_deviation => ') do
-		   		@df[0].median_absolute_deviation
+				@df[0].median_absolute_deviation
 			end
 			@result_median_absolute_deviation.append("%1.20f" % report.real)
 		end	
@@ -94,7 +96,7 @@ def self.benchmark_sum_of_squared_deviation()
 
 		bench = Benchmark.bm do |x|
 			report = x.report('Vector sum_of_squared_deviation => ') do
-		   		@df[0].sum_of_squared_deviation
+				@df[0].sum_of_squared_deviation
 			end
 			@result_sum_of_squared_deviation.append("%1.20f" % report.real)
 		end	
@@ -104,7 +106,7 @@ def self.benchmark_average_deviation_population()
 
 		bench = Benchmark.bm do |x|
 			report = x.report('Vector average_deviation_population => ') do
-		   		@df[0].average_deviation_population
+				@df[0].average_deviation_population
 			end
 			@result_average_deviation_populationa.append("%1.20f" % report.real)
 		end	
@@ -119,6 +121,14 @@ def self.benchmark_create_df(size)
 		end
 	end
 
+	def self.benchmark_unique()
+		bench = Benchmark.bm do |x|
+			report = x.report('return Unique elements => ') do
+				@df[0].uniq			end
+			@result_unique.append("%1.20f" % report.real)
+		end
+	end
+
 	def self.generate_df(size)
 		# for size * size dataframe
 		# @df= Daru::DataFrame.new(
@@ -143,10 +153,11 @@ def self.result()
 		puts 'sum_of_squared_deviation => ' , @result_sum_of_squared_deviation.join(" | ")
 		puts 'average_deviation_populationa => ' , @result_average_deviation_populationa.join(" | ")
 		puts 'create df real time => ' , @result_create_df.join(" | ")
+		puts 'Unique elements => ', @result_unique.join(" | ")
 	end
 
 	def self.result_with_size()
-		puts 'Real times for vector size [10**2, 10**3,10**4,10**5,10**6] '
+		puts 'Real times for vector size [10 ** 2, 10 ** 3,10 ** 4,10 ** 5,10 ** 6] '
 
 		self.print_array(@result_mean, 'MEAN')
 
@@ -165,6 +176,45 @@ def self.result_with_size()
 		print_array(@result_average_deviation_populationa, 'average_deviation_populationa')
 
 		print_array(@result_create_df, 'create df real time')
+		print_array(@result_unique, 'return Unique elements')
+	end
+
+	def self.result_compare()
+		puts 'Real times for vector size [10 ** 2, 10 ** 3,10 ** 4,10 ** 5,10 ** 6] '
+		puts 'Comparing with Pandas and NumPy'
+		# latest (jan 2019) Pandans and NumPy benchmark:  https://github.com/Shekharrajak/Fast-Pandas/tree/shekhar_dev
+		pandas_mean = ['0.00002460720880008011', '0.00002646757190013886', '0.00004172052699868800', '0.00015707365499838488', '0.00156584847998601610', '0.01197717989998636767', '0.12047314550000010058'] 
+		numpy_mean = ['0.00002989581900001212', '0.00003186123070008762', '0.00004564955699970596', '0.00015398673500021686', '0.00131619396001042338', '0.01254011160017398652', '0.12039573559995915553'] 
+
+		self.print_array_compare(@result_mean, 'MEAN', pandas_mean, numpy_mean)
+
+		# TODO
+		# print_array_compare(@result_mode, 'mode')
+
+		pandas_median = ['0.00002452081701998395', '0.00002761571300001378', '0.00013392778345998522', '0.00093059150948996826', '0.01047319806000086839', '0.11693990839001344728', '1.27000799899906269275'] 
+		numpy_median = ['0.00002926640069999848', '0.00003277218365001318', '0.00011260383452001407', '0.00101361991900001162', '0.00969792961999701303', '0.12528731213002175515', '1.23991956400277558714'] 
+		print_array_compare(@result_median, 'median', pandas_median, numpy_median)
+
+		pandas_sum = ['0.00005380277499971271', '0.00006364958799968007', '0.00009241857799861464', '0.00041669552999883309', '0.00496692907003307496', '0.06522532849994604198', '0.68193949160013289656'] 
+		numpy_sum = ['0.00006421745570005442', '0.00006488183500005107', '0.00012525086099776672', '0.00063505067000005507', '0.00576831682999909365', '0.07603158420024555553', '0.68379156730006795950'] 
+		print_array_compare(@result_sum, 'sum', pandas_sum, numpy_sum)
+
+		pandas_prod = ['0.00004321602870004426', '0.00004952814159987611', '0.00007920588500201119', '0.00037602675800008003', '0.00343111220001446771', '0.03398789710008713605', '0.35793153200211236253'] 
+		numpy_prod = ['0.00005185432299986132', '0.00005675383920024615', '0.00008350405899909674', '0.00039026842300154388', '0.00372708013001101781', '0.03601436539975111373', '0.35189221399923553690'] 
+		print_array_compare(@result_product, 'product', pandas_prod,  numpy_prod)
+
+		# TODO
+		# print_array_compare(@result_median_absolute_deviation, 'median_absolute_deviation')
+		# print_array_compare(@result_sum_of_squared_deviation, 'sum_of_squared_deviation')
+		# print_array_compare(@result_average_deviation_populationa, 'average_deviation_populationa')
+
+		pandas_dataframe_avg_time  =  ['0.00021143017630020040', '0.00020791667079975013', '0.00021926641200116137', '0.00021049363799829733', '0.00028836761001002739', '0.00028369350002321880', '0.00490571899717906490']
+		print_array_compare(@result_create_df, 'create df real time', pandas_dataframe_avg_time, [1,1,1,1,1,1])
+
+		pandas_unique = ['0.00003281123619999562', '0.00004384756529980222', '0.00014993138900172199', '0.00111055827999734908', '0.02723045833001378965', '0.35414526560016384993', '6.69910524100123438984'] 
+		numpy_unique = ['0.00001474316399981035', '0.00004764209610002581', '0.00059559319099935235', '0.00686007325100217689', '0.07511945111000387088', '0.90684124439976587784', '10.35867670299921883270'] 
+		print_array_compare(@result_unique, 'return Unique elements', pandas_unique, numpy_unique)
+
 	end
 
 	private
@@ -180,5 +230,16 @@ def self.print_array(array, task)
 		end
 	end
 
+	def self.print_array_compare(array, task, compare_with_array1, compare_with_array2)
+		puts 
+		puts "Method on DataFrame Vector (Vector access and apply method): **#{task}**"
+		puts
+		puts " | Number of rows | Real Time | Pandas avg time | daru/pandas | NumPy avg time | daru/numpy | "
+		puts " |------------|------------|------------|------------|------------|------------| "
+		array.each_with_index do |val, index|
+			puts " | 10 ** #{index + 2} | #{val} | #{compare_with_array1[index]} | #{Float(val)/Float(compare_with_array1[index])} | #{compare_with_array2[index]} | #{Float(val)/Float(compare_with_array2[index])} |" 
+		end
+	end
+
 end