aprendizaje_maquina 0.1.2 → 0.1.3a

checksums.yaml

@@ -1,7 +1,7 @@
 ---
 SHA1:
-  metadata.gz: 3c2e981b08fdd73029daf73b0836ced26dd40dce
-  data.tar.gz: a9e5fd556a2b44b338b670ce26b4a78fab94fa48
+  metadata.gz: 14f57cb6236d901eaef534c5ec123f3d002eae54
+  data.tar.gz: 5ace7512b5596622c67a940225d0d59209e21d2e
 SHA512:
-  metadata.gz: 2c8b5f72b79679e33dec9c26deebfabc255eb87929a2a1896b4f49e4d728cd2c4338ddc412a255a742071d199b025515132af9f67030a9408b6c125d9e32bc4c
-  data.tar.gz: 2f1066aa20a5a93879dfacd4c4557e344088cf3cd8c2168d61c64032e4e0d6d7bb327ab2f5fd844f8b46d1ad97367aa14e775c523076c1fdc3d8906ea02aca25
+  metadata.gz: c76b07f5bc6f93056138f77cace472344090a6476ecc9f5ec977caa5327f6cd1b7bf99e1bf1841051baba9f6a9e18b5f5239c146bbd59f0bca760ef14e6baf7a
+  data.tar.gz: d84bd39aa0916ae0b5125819bc2a0d64ad922a4a59d5430d8ed6a46e03a1b04cb1b9173a4654d95964eb4fef77baf557326778341a0030466166373f3e27ab4d

data/README.md

@@ -18,7 +18,7 @@ Or install it yourself as:
 
 ## Usage
 
-for make predictions with the linear regression model
+## linear regression model 
 
 first
 
@@ -27,20 +27,31 @@ first
 load data from a CSV file
 
 	load = AprendizajeMaquina::Cargar.new("file.csv")
-	y = load.to_vector(3)    # specify the column that you want to store on a vector
-    matrix = load.to_matrix  # this put all the data of the csv file in a matrix
-                             # if you don't specify the column or range of columns              
-	x = load.to_matrix(0)    # create a matrix with the data in the column 0 of the csv file 
-	                         # you can specify range like this load.to_matrix(0..4)
+
+	# specify the column that you want to store on a vector
+	y = load.to_vector(3)
+
+	# if you don't specify the column or range of columns
+	# this put all the data of the csv file in a matrix
+	matrix = load.to_matrix
+
+	# create a matrix with the data in the column 0 of the csv file       
+	x = load.to_matrix(0)    # you can specify range like this load.to_matrix(0..4)
+
 	x_with_ones = x.add_ones # this add a column of ones to the matrix
 
+to normalize data
+	
+	x.normalize
+
 create an instance of the class RegresionLineal
 
 	regresion_lineal = AprendizajeMaquina::RegresionLineal.new(x_matrix,y_vector)
-	regresion_lineal.encontrar_ecuacion    # find the theta values => Vector[114.50684133915638, 0.8310043668122375]
+	regresion_lineal.find_ecuation           # (or use the alias :train) return a Vector
+
 	m = Matrix[[1,95]]
-	p regresion_lineal.hacer_prediccion(m) # make predictions 
-										   # => Vector[193.45225618631895]
+	puts regresion_lineal.make_prediction(m) # (or use the alias :predict) to make predictions 
+					                                 # => Vector[193.45225618631895]
 
 linear regresion with arrays
 
@@ -48,14 +59,81 @@ linear regresion with arrays
 	y = [168,196,170,175,162,169,190,186,176,170,176,179]
 
 	regresion_simple = AprendizajeMaquina::RegresionLineal.new(x,y)
-	regresion_simple.encontrar_ecuacion
+	regresion_simple.train
 	p regresion_simple.ecuacion
-	p regresion_simple.hacer_prediccion(95)
+	p regresion_simple.predict(95)
+
+## Logistic Classification
+
+	data = AprendizajeMaquina::Cargar.new("data.csv")
+    
+    x = data.to_matrix(0..1).add_ones
+    y = data.to_vector(2)
+    initial_theta = Vector[0,0,0]
+    
+    cl = AprendizajeMaquina::ClasificacionLogistica.new(x,y,initial_theta)
+
+training
+
+the method ClasificacionLogistica#train receives 3 inputs, the first is the numbers of iterations, the second is the alpha value(step size), last one is type of training method ('SGD' for Stochastic Gradient Descents, 'Grad' for Batch Gradiendt Descent and 'Newm' for Newton's method)
+
+	example 1:
+	cl.train(12,0.01,'SGD')
+	example 2:
+	cl.train(10,'NewM') # Newton's method dont use alpha
+	example 3:
+	cl.train(400,0.001,'Grad')
+
+predictions
+
+	if cl.predict(Matrix[[1,24,0]]) == 1
+	  p "CANSADO"
+	else
+	  p "DESCANSADO"
+	end
+
+make predictions for multiclass(one vs all)
+
+	initial_theta_for_each_class = [ 
+	  Vector[-38.98494868465186, 3.133704064187691,-1.0058753929521247],
+		Vector[40.93814883472139,-3.2195737672278586, -0.8080682715294277],
+	  Vector[-7.220460,0.256681,1.141166]
+	]
+
+	predicted_val = []
+
+	initial_theta_for_each_class.each do |e|
+	  multiclass = AprendizajeMaquina::ClasificacionLogistica.new(x,y,e)
+	  predicted_val << multiclass.predict(Matrix[[1,13.5,1.83]])
+	end	
+
+	if predicted_val[0] == 1 
+	  puts "Vino Tinto"
+	elsif predicted_val[1] == 1
+	  puts "Vino Rosado"
+	elsif predicted_val[2] == 1
+	  puts "Vino Blanco"
+	else
+	  puts predicted_val
+	end
+
+## Clustering
+
+	load_data = AprendizajeMaquina::Cargar.new('clustering_data.csv')
+	dataset = load_data.to_matrix
+
+	# initialize with 2 cluster centroids
+	clustering = AprendizajeMaquina::KmeansClustering.new(2,dataset)
 
+	# fit the model with 20 iterations
+	clustering.fit(20)
 
-## Contributing
+	# watch the values in their respective cluster
+	p clustering.cluster(0)
+	p clustering.cluster(1)
 
-Bug reports and pull requests are welcome on GitHub at https://github.com/TheNoskOneVzla/aprendizaje_maquina.
+	# Predict the closest cluster
+	p clustering.predict(Vector[63,190])
 
 ## License
 

data/aprendizaje_maquina.gemspec

@@ -12,7 +12,6 @@ Gem::Specification.new do |spec|
 
   spec.summary       = "Machine learning gem / Una gema para el aprendizaje de maquinas."
   spec.description   = "This is a gem to help ruby developers to write machine learning algorithms easier and faster / Esta es una gema para ayudar a los desarrolladores de ruby a escribir algoritmos de aprendizaje automático más fácil y rápido."
-  spec.homepage      = ""
   spec.license       = "MIT"
 
   # Prevent pushing this gem to RubyGems.org. To allow pushes either set the 'allowed_push_host'

data/examples/clasificacion_logistica_example.rb

@@ -0,0 +1,16 @@
+require 'aprendizaje_maquina'
+
+data = AprendizajeMaquina::Cargar.new("data_lg.csv")
+x = data.to_matrix(0..1).add_ones
+y = data.to_vector(2)
+theta = Vector[0,0,0]
+rl = AprendizajeMaquina::ClasificacionLogistica.new(x,y,theta)
+
+rl.train(12,0.01,'SGD')
+prediction = rl.predict(Matrix[[1,9,22]])
+
+if prediction == 1
+	puts "cansado"
+else
+	puts "descansado"
+end

data/examples/clustering_data.csv

@@ -0,0 +1,177 @@
+63, 190
+65, 145
+69, 160
+66, 130
+63, 120
+65, 125
+68, 124
+64, 135
+64, 98
+66, 150
+67, 160
+67, 140
+71, 170
+60, 109
+68, 170
+63, 165
+67, 147
+63, 110
+68, 160
+60, 125
+66, 142
+65, 154
+62, 119
+65, 135
+69, 195
+63, 110
+63, 140
+64, 219
+66, 135
+66, 140
+68, 135
+68, 99
+64, 100
+62, 135
+74, 200
+63, 150
+69, 163
+71, 116
+68, 145
+71, 140
+68, 125
+66, 130
+72, 157
+68, 200
+61, 110
+64, 165
+70, 115
+67, 153
+65, 135
+62, 116
+63, 122
+66, 130
+73, 180
+66, 110
+69, 190
+68, 160
+62, 115
+66, 130
+64, 110
+70, 150
+72, 185
+70, 175
+66, 200
+68, 180
+62, 135
+63, 120
+63, 134
+72, 185
+72, 160
+68, 160
+65, 143
+68, 180
+65, 130
+65, 200
+65, 195
+66, 160
+70, 195
+63, 120
+69, 120
+64, 110
+67, 140
+66, 104
+63, 125
+71, 140
+68, 190
+65, 125
+66, 125
+64, 125
+72, 245
+74, 235
+62, 130
+68, 135
+62, 120
+63, 155
+65, 130
+64, 130
+74, 200
+70, 145
+68, 140
+69, 135
+61, 124
+66, 125
+76, 216
+66, 120
+62, 103
+65, 162
+69, 165
+73, 164
+70, 153
+75, 172
+58, 103
+63, 135
+68, 160
+69, 141
+67, 145
+68, 148
+67, 180
+70, 150
+70, 160
+60, 170
+70, 210
+66, 175
+64, 130
+72, 172
+64, 108
+72, 190
+66, 142
+69, 235
+72, 145
+66, 143
+70, 141
+64, 110
+70, 196
+65, 118
+72, 200
+61, 145
+63, 110
+65, 115
+70, 170
+66, 160
+62, 123
+65, 135
+68, 140
+65, 160
+60, 150
+72, 168
+61, 100
+62, 100
+66, 135
+67, 155
+62, 135
+64, 200
+66, 140
+62, 185
+65, 125
+72, 165
+64, 95
+72, 155
+65, 110
+67, 165
+65, 132
+66, 125
+64, 120
+63, 130
+66, 130
+63, 115
+59, 160
+64, 150
+64, 130
+61, 120
+66, 150
+66, 250
+71, 150
+70, 180
+69, 210
+61, 105
+64, 140

data/examples/clustering_example.rb

@@ -0,0 +1,9 @@
+require 'aprendizaje_maquina'
+
+load_data = AprendizajeMaquina::Cargar.new('clustering_data.csv')
+dataset = load_data.to_matrix
+clustering = AprendizajeMaquina::KmeansClustering.new(2,dataset)
+clustering.fit(20)
+p clustering.cluster(0)
+p clustering.cluster(1)
+p clustering.predict(Vector[63,190])

data/examples/data_lg.csv

@@ -0,0 +1,145 @@
+0,24,1
+0.1,23.5,1
+0.2,23.4,1
+0.3,23.3,1
+0.4,23.2,1
+0.5,23.1,1
+1,23,1
+1.1,23.5,1
+1.2,23.4,1
+1.3,23.3,1
+1.4,23.2,1
+1.5,23.1,1
+2,22,1
+2.1,22.5,1
+2.2,22.4,1
+2.3,22.3,1
+2.4,22.2,1
+2.5,22.1,1
+3,21,1
+3.1,21.5,1
+3.2,21.4,1
+3.3,21.3,1
+3.4,21.2,1
+3.5,21.1,1
+4,20,1
+4.1,20.5,1
+4.2,20.4,1
+4.3,20.3,1
+4.4,20.2,1
+4.5,20.1,1
+5,19,1
+5.1,19.5,1
+5.2,19.4,1
+5.3,19.3,1
+5.4,19.2,1
+5.5,19.1,1
+6,18,1
+6.1,18.5,1
+6.2,18.4,1
+6.3,18.3,1
+6.4,18.2,1
+6.5,18.1,1
+7,17,1
+7.1,17.5,1
+7.2,17.4,1
+7.3,17.3,1
+7.4,17.2,1
+7.5,17.1,1
+8,16,0
+8.1,16.5,0
+8.2,16.4,0
+8.3,16.3,0
+8.4,16.2,0
+8.5,16.1,0
+9,15,0
+9.1,15.5,0
+9.2,15.4,0
+9.3,15.3,0
+9.4,15.2,0
+9.5,15.1,0
+10,14,0
+10.1,14.5,0
+10.2,14.4,0
+10.3,14.3,0
+10.4,14.2,0
+10.5,14.1,0
+11,13,0
+11.1,13.5,0
+11.2,13.4,0
+11.3,13.3,0
+11.4,13.2,0
+11.5,13.1,0
+12,12,0
+12.1,12.5,0
+12.2,12.4,0
+12.3,12.3,0
+12.4,12.2,0
+12.5,12.1,0
+13,11,0
+13.1,11.5,0
+13.2,11.4,0
+13.3,11.3,0
+13.4,11.2,0
+13.5,11.1,0
+14,10,0
+14.1,10.5,0
+14.2,10.4,0
+14.3,10.3,0
+14.4,10.2,0
+14.5,10.1,0
+15,9,0
+15.1,9.5,0
+15.2,9.4,0
+15.3,9.3,0
+15.4,9.2,0
+15.5,9.1,0
+16,8,0
+16.1,8.5,0
+16.2,8.4,0
+16.3,8.3,0
+16.4,8.2,0
+16.5,8.1,0
+17,7,0
+17.1,7.5,0
+17.2,7.4,0
+17.3,7.3,0
+17.4,7.2,0
+17.5,7.1,0
+18,6,0
+18.1,6.5,0
+18.2,6.4,0
+18.3,6.3,0
+18.4,6.2,0
+18.5,6.1,0
+19,5,0
+19.1,5.5,0
+19.2,5.4,0
+19.3,5.3,0
+19.4,5.2,0
+19.5,5.1,0
+20,4,0
+20.1,4.5,0
+20.2,4.4,0
+20.3,4.3,0
+20.4,4.2,0
+20.5,4.1,0
+21,3,0
+21.1,3.5,0
+21.2,3.4,0
+21.3,3.3,0
+21.4,3.2,0
+21.5,3.1,0
+22,2,0
+22.1,2.5,0
+22.2,2.4,0
+22.3,2.3,0
+22.4,2.2,0
+22.5,2.1,0
+23,1,0
+23.1,1.5,0
+23.2,1.4,0
+23.3,1.3,0
+23.4,1.2,0
+23.5,1.1,0
+24,0,0