bias-variance

Author: Lewuathe

.cache

@@ -10,4 +10,6 @@ project/boot/
 project/plugins/project/
 
 # Scala-IDE specific
-.scala_dependencies
+.scala_dependencies
+
+*.csv

.gitignore

@@ -0,0 +1,15 @@
+# -*- coding: utf-8 -*-
+
+import os
+import sys
+import numpy as np
+
+if __name__ == "__main__":
+    for k in range(0, 20):
+        xlist = np.linspace(0, 1, 25)
+        tlist = np.sin(2*np.pi*xlist) + np.random.normal(0, 0.2, xlist.size)
+        f = open("train" + str(k) + ".csv", "w")
+        for i in range(0, len(xlist)):
+            f.write(str(xlist[i]) + "," + str(tlist[i]) + "\n")
+
+        f.close()

estimate.csv

@@ -7,18 +7,11 @@ import com.github.tototoshi.csv._
 
 object App {
   def main(args: Array[String]) {
-    //val data = EmployeeData.read()
+    /*
     val train = DataReader.trainRead("train.csv")
     
-    //val data = List[Pair[Double,Double]]((0.07218849, 1.47822651), (1.61854596, 11.57308266), (2.74441742, 23.53791146), (3.36337206, 36.71027063), (4.96067527, 40.60451491), (5.13515737, 58.41530339), ( 6.43154354, 64.22937428), (7.68320999, 79.51190066), (8.3692042, 91.04207353), ( 9.85696038, 97.7546571))
-    //val data = List[Pair[Double,Double]]((1.0,1.0),(2.0,2.0),(3.0,3.0))
-    
-    //val data = List[Pair[Double,Double]]((1.0, Math.sin(1.0)), (2.0, Math.sin(2.0)), (3.0, Math.sin(3.0)))
-    val wlist = LeastSquares.estimate(train, 0.0)
-    val test = DataReader.testRead("test.csv")
-    
-    
     val estimate = new PrintWriter("estimate.csv")
+    
    	val sums = DenseMatrix.zeros[Double](BayesEstimation.M+1, BayesEstimation.M+1)
     for(i <- 0 to train.length-1) {
 	  sums += BayesEstimation.phi(train(i)._1) * BayesEstimation.phi(train(i)._1).t
@@ -36,18 +29,25 @@ object App {
     	  estimate.println(x + "," + m + "," + u + "," + l)
       }
     }
-    
-    /*
-    test.foreach {
-      x => {
-        estimate.println(x + "," + LinearEquation.y(x, wlist))
-      }
-    }
     * 
     */
-	
-    estimate.flush()
-    estimate.close()
+
+    for(i <- 0 to 19) {
+    	val tfilename = "train" + i + ".csv"
+    	val efilename = "estimate" + i + ".csv"
+    	val train = DataReader.trainRead(tfilename)
+    	val wlist = LeastSquares.estimate(train, 10.0)
+    	val test  = DataReader.testRead("test.csv")
+    	val estimate = new PrintWriter(efilename)
+    
+    	test.foreach {
+    		x => {
+    			estimate.println(x + "," + LinearEquation.y(x, wlist))
+    		}
+    	}
+    	estimate.flush()
+    	estimate.close()
+    }
 
 
   }

genTrain.py

@@ -1,20 +0,0 @@
-0.0,-0.3064923 
-0.05263158,0.57885623
-0.10526316,0.62210423
-0.15789474,0.68922288
-0.21052632,1.10374508
-0.26315789,0.98969203
-0.31578947,1.30612747
-0.36842105,1.00316323
-0.42105263,0.6029487
-0.47368421,0.04878376
-0.52631579,-0.49498364
-0.57894737,-0.87668114
-0.63157895,-0.65494625
-0.68421053,-1.10763935
-0.73684211,-1.13152455
-0.78947368,-0.72751465
-0.84210526,-0.98385029
-0.89473684,-0.755865
-0.94736842,-0.47920383
-1.0,-0.0789461      

results/lambda0.0000000001.png

@@ -0,0 +1,31 @@
+# -*- coding: utf-8 -*-
+
+import os
+import sys
+import csv
+from  numpy import *
+from pylab import *
+
+if __name__ == "__main__":
+    for i in range(0, 19):
+        estimate = open('estimate' + str(i) + '.csv', 'rb')
+            
+        estreader = csv.reader(estimate)
+        estxs = []
+        estts = []
+        for row in estreader:
+            estxs.append(row[0])
+            estts.append(row[1])
+        estxs = array(estxs)
+        estts = array(estts)
+                
+                
+#        plot(trainxs, traints, 'bo')
+        plot(estxs, estts, 'g-')
+        xlim(0.0, 1.0)   
+        ylim(-1.5, 1.5)
+        estimate.close()
+        
+    show()  
+        
+