Change@abhishek

Readme.md

@@ -1,4 +1,3 @@
-
 # Qikify
 
   Qikify is an MVC framework for adaptive test and semiconductor data analysis
@@ -43,6 +42,7 @@ TODO: Need example demonstrating usage.
 The following are the major project contributors.
 
   * Nate Kupp ([natekupp](https://github.com/natekupp))
+  * Abhishek Basu([abhishekingithub](https://github.com/abhishekingithub))
 
 ## License 
 

algorithm_implmentation/dataset.txt

@@ -0,0 +1,12 @@
+42.8  40.0  37
+63.5  93.5  50
+37.5  35.5  34
+39.5  30.0  36
+45.5  52.0  43
+38.5  17.0  28
+43.0  38.5  37
+22.5   8.5  20
+37.0  33.0  34
+23.5   9.5  30
+33.0  21.0  38
+58.0  79.0  47

algorithm_implmentation/linearregression_gradientdescentmodel.py

@@ -0,0 +1,122 @@
+import numpy as np
+import matplotlib.pyplot as plt
+
+class linearregression_gradientdescentmodel(object):
+	def __init__(self):
+		self.chip_theta = []
+		self.chip_targetvalues = []
+		self.sumofx_i=0
+		self.alpha=1e-6
+		self.m=0
+		#self.rows=list()
+		#self.thetas=list()
+		self.estimated_theta=list()
+		#self.inittheta0=0
+		self.no_features=0
+		self.feature_matrix=[[]]
+		self.int_feature_matrix=[[]]
+		self.sum=0
+		self.n1=list()
+		self.n2=list()
+
+	def form_hypothesis(self):
+		init_thetas=list()
+		datafile = open("dataset.txt",'r')
+		for line in datafile:
+			line=line.split("\n")[0]
+			self.m+=1
+			featurelist=line.split('  ')
+			self.no_features=len(featurelist)
+
+			print "featurelength %i" %self.no_features
+			if self.m==1:
+				self.feature_matrix[0]=featurelist
+			else:
+				self.feature_matrix.append(featurelist)
+		print self.feature_matrix
+		print self.m
+		featurelength=len(featurelist)
+		self.int_feature_matrix=np.array(self.feature_matrix,dtype=float)
+		#self.sum_y=self.calculate_sum_colomn(self.int_feature_matrix,len(featurelist)-1)
+		for j in range(0,len(featurelist)):
+			random_theta=np.random.random()
+			self.estimated_theta.append(random_theta)
+
+		for n in range(0,1000):
+			#coloumn_list=list()
+			sum_coloumn=list()
+		#	hypo_sum=0
+		#	linear_sum=0
+		#	calsum=0
+		#	iteration=0	
+
+			for j in range(0,len(featurelist)):
+				hypo_sum=0
+				linear_sum=0
+				calsum=0
+				diff=0		
+				for y in range(0,self.m):
+					#if y==0:
+					value_x=list()
+					for x in range(0,len(featurelist)):
+						hypo_sum=self.calculate_sum(calsum,self.estimated_theta,self.int_feature_matrix,y,x)
+						#coloumn_list+str(x)=list()
+						#coloumn_list+str(x)=self.int_feature_matrix[:,x]
+					diff+=pow(hypo_sum-self.int_feature_matrix[y][featurelength-1],2)
+					#print 'cost value'
+					#print cost
+					#print 1/float(self.m)
+					if j==0:
+						linear_sum+=(hypo_sum-self.int_feature_matrix[y][featurelength-1])
+					else:
+						#name=coloumn_list+str(j)
+						#print name
+						linear_sum+=(hypo_sum-self.int_feature_matrix[y][featurelength-1])*self.int_feature_matrix[y][j]
+					#print self.int_feature_matrix[y][featurelength-1]
+				sum_coloumn.append(linear_sum)
+			cost=(1/float(2*self.m))*diff
+			print 'cost value..........'
+			print cost
+			self.n1.append(n)
+			self.n2.append(cost)
+			costlength=len(self.n2)
+			n3=self.n2[costlength-2]
+			#print n3
+			if costlength>1 and cost>n3:
+				#val=self.alpha
+				self.alpha=self.alpha/float(10)
+			#plt.show()
+			#print linear_sum
+			for k in range(0,featurelength):
+				temp=self.estimated_theta[k]-self.alpha*(1/float(self.m))*sum_coloumn[k]
+				self.estimated_theta.remove(self.estimated_theta[k])
+				self.estimated_theta.insert(k,temp)
+
+			print self.estimated_theta
+		plt.plot(self.n1,self.n2)
+		plt.show()
+	def calculate_sum(self,hypo_sum,thetas,feature_matrix,y,x):
+		if x==0:
+			#sum=0
+			hypo_sum=thetas[x]
+			#htheta=htheta+thetas[x+1]*float(feature_matrix[y][x])
+		else:
+			#print "y value %f"%feature_matrix[y][featurelength-1]
+			hypo_sum+=thetas[x]*self.int_feature_matrix[y][x-1]
+		#rows.append(featurelist[x])
+		return hypo_sum
+
+	def calculate_sum_colomn(self,matrix,i):
+		#sumofx=0
+		coloumn_list=list()
+		print 'in sum............'
+		print i
+		#for i in range (0,self.m-1):
+		#import pdb; pdb.set_trace()
+		#import pdb; pdb.set_trace()
+		coloumn_list=matrix[:,i]
+		return np.sum(np.array(coloumn_list))
+
+if __name__ == "__main__":
+	l_reg=linearregression_gradientdescentmodel()
+	l_reg.form_hypothesis()

qikify/models/test_knn.py

@@ -0,0 +1,14 @@
+from qikify.models.KNNImplementation import KNNImplementation
+from qikify.models import Chip
+def test_knn():
+	knn=KNNImplementation()
+	chip_data1={'ORB_a':1,'ORB_b':1,'gnd':1}
+	chip_data2={'ORB_a':-1,'ORB_b':-1,'gnd':-1}
+	chip1=Chip(chip_data1,LCT_prefix='ORB')
+	chip2=Chip(chip_data2,LCT_prefix='ORB')
+	chips=[chip1,chip2]
+	print "hi"
+	knn.trainmodel(chips)
+
+if __name__ == "__main__":
+	test_knn()