def Test2():
dataSet = read_dataset.DataSet('./DataSet/Exams_grad.csv')
learningRate = 0.5
numofIteration = 500
samples_values_arr, target_feature_arr = dataSet.getDataSetAsNumPy()
target_feature_arr = pd.DataFrame(target_feature_arr)
logReg = LogisticRegression(samples_values_arr, target_feature_arr)
model_weights, regressionLine = logReg.gradientDescent(initlearningRate=learningRate,
numofIteration=numofIteration)
plt.figure(1)
plt.scatter(samples_values_arr.iloc[:,0], samples_values_arr.iloc[:,1], s=10 * target_feature_arr,
label='faulty', marker='x')
plt.scatter(samples_values_arr.iloc[:,0], samples_values_arr.iloc[:,1], s=10 * (1 - target_feature_arr),
label='good', marker='o')
plt.xlabel(dataSet.dataSet.columns.values[0])
plt.ylabel(dataSet.dataSet.columns.values[1])
print(regressionLine)
plt.plot(samples_values_arr.iloc[:,0], regressionLine, color='r')
plt.title(dataSet.data_set_name + ' data set')
# print(logReg.square_error)
#
plt.figure(2)
plt.plot(range(numofIteration), logReg.square_error)
plt.xlabel('Num of iterations')
plt.ylabel('Square Error')
plt.title('Cost Function over iteration')
plt.legend()
plt.show()
def Test1():
dataSet = read_dataset.DataSet('./DataSet/franchise_rest.csv')
learningRate = 0.5
numofIteration = 100
samples_values_arr, target_feature_arr = dataSet.getDataSetAsNumPy()
linearReg = LinearRegression(samples_values_arr, target_feature_arr)
model_weights, regressionLine = linearReg.gradientDescent(
initlearningRate=learningRate, numofIteration=numofIteration)
plt.figure(1)
plt.scatter(samples_values_arr,
target_feature_arr,
label=dataSet.data_set_name,
marker='x')
plt.xlabel(dataSet.dataSet.columns.values[0])
plt.ylabel(dataSet.dataSet.columns.values[-1])
plt.plot(samples_values_arr, regressionLine)
plt.title(dataSet.data_set_name + ' data set')
print(linearReg.square_error)
plt.figure(2)
plt.plot(range(numofIteration), linearReg.square_error)
plt.xlabel('Num of iterations')
plt.ylabel('Square Error')
plt.title('Cost Function over iteration')
plt.legend()
plt.show()
def Test3():
dataSet = read_dataset.DataSet('./DataSet/House_Price.csv')
learningRate = 1.5
numofIteration = 100
samples_values_arr, target_feature_arr = dataSet.getDataSetAsNumPy()
linearReg = LinearRegression(samples_values_arr, target_feature_arr)
model_weights, regressionLine = linearReg.gradientDescent(
initlearningRate=learningRate, numofIteration=numofIteration)
t_stat, p_values, null_hypothesis = linearReg.stat_significance_calc()
print('t-statistic :')
print(t_stat)
print('P_values :')
print(p_values)
print('significant impact')
print(null_hypothesis)
fig = plt.figure()
ax = Axes3D(fig)
ax.scatter3D(samples_values_arr.iloc[:, 0],
samples_values_arr.iloc[:, 1],
np.transpose(target_feature_arr),
marker='x',
label="The actual Values")
ax.scatter3D(samples_values_arr.iloc[:, 0],
samples_values_arr.iloc[:, 1],
regressionLine,
marker='o',
label="The predicated Values")
ax.set_xlabel(dataSet.dataSet.columns.values[0])
ax.set_ylabel(dataSet.dataSet.columns.values[1])
ax.set_zlabel(dataSet.dataSet.columns.values[2])
plt.title(dataSet.data_set_name + ' data set')
print(linearReg.square_error)
# #
plt.figure(2)
plt.plot(range(numofIteration), linearReg.square_error)
plt.xlabel('Num of iterations')
plt.ylabel('Square Error')
plt.title('Cost Function over iteration')
#
# plt.legend()
plt.show()