from sklearn.metrics import confusion_matrix import anomalydensity as an #user defined module #reading csv with pandas as np.loadtxt throws error if column names are present xtrain = pd.read_csv('xtrain.csv',delimiter=',') #1000*11 training set xtest = pd.read_csv('xtest.csv',delimiter=',') #1000*11 testing set ytest = pd.read_csv('ytest.csv',delimiter=',') #1000*1 original y values #converting from pandas dataframe to numpy array xtrain1 =np.array(xtrain) xtest1 = np.array(xtest) ytest1 = np.array(ytest) #creating object1 b = an.anomaly(xtrain1) #anomaly is a class inside anomalydensity module par = b.parameter() p_density=b.multivariateGaussian(par[0],par[2]) #creating object2 c=an.anomaly(xtest1) pval_density=c.multivariateGaussian(par[0],par[2]) print p_density.shape,pval_density.shape type(p_density),type(pval_density) epsilon = np.mean(pval_density) bestepsilon = 0. bestf1=0. f1=0. pred=np.empty([1000,])
#converting from pandas dataframe to numpy array xtrain2 =np.array(xtrain_s) xtest2 = np.array(xtest_s) ytest2 = np.array(ytest_s) matplotlib.rcParams['xtick.direction'] = 'out' matplotlib.rcParams['ytick.direction'] = 'out' #Initializing the plot xplt = np.arange(0, 36, 0.5) yplt = np.arange(0,36,0.5) X, Y = np.meshgrid(xplt, yplt) H = np.column_stack((X.flatten(1),Y.flatten(1))) #creating object1 b1 = an.anomaly(xtrain2) #anomaly is a class inside anomalydensity module par1 = b1.parameter() p_density1=b1.multivariateGaussian(par1[0],par1[2]) #creating object for plot plt1 = an.anomaly(H) #anomaly is a class inside anomalydensity module plt_density=plt1.multivariateGaussian(par1[0],par1[2]) plt_density.resize((72,72)) plt_density.shape levels = [1.0000e-020,1.0000e-017,1.0000e-014,1.0000e-011,1.0000e-008,1.0000e-005,1.0000e-002] fig=plt.figure(figsize=(30,20)) ax1 = fig.add_subplot(121) ax1.scatter((xtrain2[:,0]),(xtrain2[:,1]),color='blue',s=5,edgecolor='none') ax1.set_xlabel('Latency') ax1.set_ylabel('Throughput')
import scipy as sp from sklearn.metrics import confusion_matrix import anomalydensity as an #user defined module #reading csv with pandas as np.loadtxt throws error if column names are present xtrain = pd.read_csv('xtrain.csv', delimiter=',') #1000*11 training set xtest = pd.read_csv('xtest.csv', delimiter=',') #1000*11 testing set ytest = pd.read_csv('ytest.csv', delimiter=',') #1000*1 original y values #converting from pandas dataframe to numpy array xtrain1 = np.array(xtrain) xtest1 = np.array(xtest) ytest1 = np.array(ytest) #creating object1 b = an.anomaly(xtrain1) #anomaly is a class inside anomalydensity module par = b.parameter() p_density = b.multivariateGaussian(par[0], par[2]) #creating object2 c = an.anomaly(xtest1) pval_density = c.multivariateGaussian(par[0], par[2]) print p_density.shape, pval_density.shape type(p_density), type(pval_density) epsilon = np.mean(pval_density) bestepsilon = 0. bestf1 = 0. f1 = 0. pred = np.empty([