CNTK - Simple linear regression

The following are sample code to create a linear regression to work with car data.

	import pandas as pd
	import numpy as np
	import cntk as C
	trainData = pd.read_csv("cardata.csv")
	testData = pd.read_csv("testCarData.csv")
	#print(trainData)
	#print(testData)
	input_dim = 2
	num_output_classes = 2
	#convert data => 0 1 => if it is a turbo
	class_ind = [trainData['IsRaceCar'] == class_number for class_number in range(2)]
	#print('****')
	#print(class_ind)
	plotLabel = np.asarray(np.hstack(class_ind), dtype=np.float32)
	# We expect our data features to be in [[1,2], [2,1]] ... format
	def getData(targetData):
	alist = []
	for idx in range(0, targetData['CarCC'].size):
	idxVal = [targetData['CarCC'].values[idx], targetData['Turbo'].values[idx]]
	alist.append(idxVal)
	labels = []
	for class_number in trainData['IsRaceCar']:
	raceValue = class_number == 1
	if raceValue == 1:
	nraceValue = 0
	elif raceValue == 0:
	nraceValue = 1
	a = [class_number, nraceValue]
	labels.append(a)
	return alist, labels
	dataFeatures, labels = getData(trainData)
	print('train data samples')
	print(dataFeatures)
	print(labels)
	import matplotlib.pyplot as plt
	#print(plotLabel)
	colors = ['r' if label == 0 else 'b' for label in plotLabel]
	plt.scatter(trainData['CarCC'], trainData['Turbo'], c=colors)
	plt.xlabel("Engine size")
	plt.ylabel("With Turbo")
	plt.show()
	## starting off with a not so complicated linear model
	mydict = {}
	def linear_layer(input_var, output_dim):
	input_dim = input_var.shape[0]
	weight_param = C.parameter(shape=(input_dim, output_dim))
	bias_param = C.parameter(shape=(output_dim))
	mydict['w'], mydict['b'] = weight_param, bias_param
	return C.times(input_var, weight_param) + bias_param
	feature = C.input_variable(input_dim, np.float32)
	#print(feature)
	output_dim = num_output_classes
	z = linear_layer(feature, output_dim)
	## pretty standard stuff ###
	progress_printer = C.logging.ProgressPrinter(tag='Training', num_epochs=25)
	C.logging.log_number_of_parameters(z)
	label = C.input_variable(num_output_classes, np.float32)
	loss = C.cross_entropy_with_softmax(z, label)
	eval_error = C.classification_error(z, label)
	learning_rate = 0.5
	lr_schedule = C.learning_rate_schedule(learning_rate, C.UnitType.minibatch)
	learner = C.sgd(z.parameters, lr_schedule)
	trainer = C.Trainer(z, (loss, eval_error), [learner], progress_printer)
	minibatch_size = 25
	for i in range(0, 5):
	# Assign the minibatch data to the input variables and train the model on the minibatch
	trainer.train_minibatch({feature : dataFeatures, label : labels})
	dataFeatures, labels = getData(testData)
	trainer.test_minibatch({feature : dataFeatures, label : labels})
	predDataFeature = [[2500,1], [300,0]]
	out = C.softmax(z)
	result = out.eval({feature : predDataFeature})
	print('predication results:')
	print(result)
	print("Predicted:", [np.argmax(x) for x in result])
	CarData.csv + TestCarData.csv
	CarCC,Turbo,IsRaceCar
	3000,1,1
	2500,1,1
	1300,0,0
	1200,0,0
	1600,0,0
	2000,0,0
	1800,0,0
	1500,0,0
	900,0,0
	700,0,0
	800,0,0
	700,0,0
	1100,0,0
	1300,0,0
	1200,0,0
	1600,0,0
	2000,0,0
	1800,0,0
	1500,0,0
	900,0,0
	700,0,0
	800,0,0
	700,0,0
	1100,0,0
	1300,0,0
	1200,0,0
	1600,0,0
	2000,0,0
	1800,0,0
	1500,0,0
	900,0,0
	700,0,0
	800,0,0
	700,0,0
	1100,0,0

view raw linear_regression_cntk_car.py hosted with ❤ by GitHub

	import pandas as pd
	import numpy as np
	import cntk as C
	trainData = pd.read_csv("cardata.csv")
	testData = pd.read_csv("testCarData.csv")
	#print(trainData)
	#print(testData)
	input_dim = 2
	num_output_classes = 2
	#convert data => 0 1 => if it is a turbo
	class_ind = [trainData['IsRaceCar'] == class_number for class_number in range(2)]
	#print('****')
	#print(class_ind)
	plotLabel = np.asarray(np.hstack(class_ind), dtype=np.float32)
	# We expect our data features to be in [[1,2], [2,1]] ... format
	def getData(targetData):
	alist = []
	for idx in range(0, targetData['CarCC'].size):
	idxVal = [targetData['CarCC'].values[idx], targetData['Turbo'].values[idx]]
	alist.append(idxVal)
	labels = []
	for class_number in trainData['IsRaceCar']:
	raceValue = class_number == 1
	if raceValue == 1:
	nraceValue = 0
	elif raceValue == 0:
	nraceValue = 1
	a = [class_number, nraceValue]
	labels.append(a)
	return alist, labels
	dataFeatures, labels = getData(trainData)
	print('train data samples')
	print(dataFeatures)
	print(labels)
	import matplotlib.pyplot as plt
	#print(plotLabel)
	colors = ['r' if label == 0 else 'b' for label in plotLabel]
	plt.scatter(trainData['CarCC'], trainData['Turbo'], c=colors)
	plt.xlabel("Engine size")
	plt.ylabel("With Turbo")
	plt.show()
	## starting off with a not so complicated linear model
	mydict = {}
	def linear_layer(input_var, output_dim):
	input_dim = input_var.shape[0]
	weight_param = C.parameter(shape=(input_dim, output_dim))
	bias_param = C.parameter(shape=(output_dim))
	mydict['w'], mydict['b'] = weight_param, bias_param
	return C.times(input_var, weight_param) + bias_param
	feature = C.input_variable(input_dim, np.float32)
	#print(feature)
	output_dim = num_output_classes
	z = linear_layer(feature, output_dim)
	## pretty standard stuff ###
	progress_printer = C.logging.ProgressPrinter(tag='Training', num_epochs=25)
	C.logging.log_number_of_parameters(z)
	label = C.input_variable(num_output_classes, np.float32)
	loss = C.cross_entropy_with_softmax(z, label)
	eval_error = C.classification_error(z, label)
	learning_rate = 0.5
	lr_schedule = C.learning_rate_schedule(learning_rate, C.UnitType.minibatch)
	learner = C.sgd(z.parameters, lr_schedule)
	trainer = C.Trainer(z, (loss, eval_error), [learner], progress_printer)
	minibatch_size = 25
	for i in range(0, 5):
	# Assign the minibatch data to the input variables and train the model on the minibatch
	trainer.train_minibatch({feature : dataFeatures, label : labels})
	dataFeatures, labels = getData(testData)
	trainer.test_minibatch({feature : dataFeatures, label : labels})
	predDataFeature = [[2500,1], [300,0]]
	out = C.softmax(z)
	result = out.eval({feature : predDataFeature})
	print('predication results:')
	print(result)
	print("Predicted:", [np.argmax(x) for x in result])
	CarData.csv + TestCarData.csv
	CarCC,Turbo,IsRaceCar
	3000,1,1
	2500,1,1
	1300,0,0
	1200,0,0
	1600,0,0
	2000,0,0
	1800,0,0
	1500,0,0
	900,0,0
	700,0,0
	800,0,0
	700,0,0
	1100,0,0
	1300,0,0
	1200,0,0
	1600,0,0
	2000,0,0
	1800,0,0
	1500,0,0
	900,0,0
	700,0,0
	800,0,0
	700,0,0
	1100,0,0
	1300,0,0
	1200,0,0
	1600,0,0
	2000,0,0
	1800,0,0
	1500,0,0
	900,0,0
	700,0,0
	800,0,0
	700,0,0
	1100,0,0

view raw linear_regression_cntk_car.py hosted with ❤ by GitHub