def drawSimilarities(placeOfWordToCompare):
plt.figure()
file = importation.importcsv()
k = 0
for line in file:
value = float(line[placeOfWordToCompare]) * 100
#print(value)
if int(line[-1]) == 1:
color = 'blue'
if value < 200:
plt.plot(value,
value + 10,
color=color,
linestyle='dashed',
marker='o',
markerfacecolor=color,
markersize=3)
else:
color = 'orange'
if value < 200:
plt.plot(value,
value - 10,
color=color,
linestyle='dashed',
marker='o',
markerfacecolor=color,
markersize=3)
k += 1
print('rrrr')
plt.show()
KNeighborsClassifier(n_neighbors = 2, weights = "distance"),
#GaussianProcessClassifier(1.0 * RBF(1.0)),
#GaussianNB(),
SVC(gamma=2, C=1)
#QuadraticDiscriminantAnalysis(),
]
names = ["AdaBoost", "Linear SVM", "Neural Net", "Decision Tree", "Random Forest", "Nearest Neighbors",
"RBF SVM"
]#"QDA",#"Gaussian Process","Naive Bayes"
rowData = importcsv("spambase.data")
data = []
for line in rowData:
listLine = []
for value in line:
listLine.append(float(value))
data.append(listLine)
usedData = []
usedValue = []
for line in data:
listLine = []
for k in range(len(line)):
if k not in [27, 28, 31, 57, 0,3,14,16,22,24,26,30,31, 32, 33, 34, 37, 39, 40, 41,42, 46,47, 50, 51]:
listLine.append(line[k])
def drawSimilarities(listOfPlace):
plt.figure()
file = importation.importcsv()
k = 0
totSpam = 0
totNonSpam = 0
for line in file:
value = float(line[placeOfWordToCompare])
if value < 0.05:
if int(line[-1]) == 1:
spamClasses["<0.05"].append(value)
totSpam += 1
else:
nonSpamClasses["<0.05"].append(value)
totNonSpam += 1
elif 0.05 <= value and value < 0.5:
if int(line[-1]) == 1:
spamClasses["0.05<0.5"].append(value)
totSpam += 1
else:
nonSpamClasses["0.05<0.5"].append(value)
totNonSpam += 1
elif 0.5 <= value and value < 1:
if int(line[-1]) == 1:
spamClasses["0.5<1"].append(value)
totSpam += 1
else:
nonSpamClasses["0.5<1"].append(value)
totNonSpam += 1
elif 1 <= value and value < 1.5:
if int(line[-1]) == 1:
spamClasses["1<1.5"].append(value)
totSpam += 1
else:
nonSpamClasses["1<1.5"].append(value)
totNonSpam += 1
elif 1.5 <= value and value < 2:
if int(line[-1]) == 1:
spamClasses["1.5<2"].append(value)
totSpam += 1
else:
nonSpamClasses["1.5<2"].append(value)
totNonSpam += 1
elif 2 <= value:
if int(line[-1]) == 1:
spamClasses["2<"].append(value)
totSpam += 1
else:
nonSpamClasses["2<"].append(value)
totNonSpam += 1
nb = 0
for value in spamClasses:
tot = (len(spamClasses[value]) / totSpam) * 100
plt.plot(nb,
tot,
color='blue',
linestyle='dashed',
marker='o',
markerfacecolor='blue',
markersize=5)
nb += 1
print("\n")
print(value)
print(tot)
print(totSpam)
nb = 0
print("\n non spam : \n")
for value in nonSpamClasses:
tot = (len(nonSpamClasses[value]) / totNonSpam) * 100
plt.plot(nb,
tot,
color='orange',
linestyle='dashed',
marker='o',
markerfacecolor='orange',
markersize=5)
nb += 1
print("\n")
print(value)
print(tot)
print(totNonSpam)
print('rrrr')
plt.show()
from sklearn.naive_bayes import GaussianNB
from random import random
from importation import importcsv
import numpy as np
# Load data
rowData = importcsv()
data = []
for line in rowData:
listLine = []
for value in line:
listLine.append(float(value))
data.append(listLine)
usedData = []
usedValue = []
for line in data:
listLine = []
for k in range(len(line)):
if k != 27 and k != 28 and k != 31 and k != 57:
listLine.append(line[k])
usedValue.append(line[-1])
usedData.append(listLine)
'''data = np.array(data)
print(data)
'''
testSetX = []
testSetY = []
for k in range(1000):
placeToTakeForTest = int(random() * len(usedData))
def drawSimilarities(listPlaceOfWordToCompare,
precision,
valueToCompare,
pathFile,
draw=True):
nbOk = []
file = importcsv(pathFile)
for placeOfWordToCompare in listPlaceOfWordToCompare:
#print("place of word : ")
#print(placeOfWordToCompare)
if draw:
plt.figure()
plt.title('Word nb = ' + str(placeOfWordToCompare))
k = 0
spamClasses = {"<value": [], "value<": []}
nonSpamClasses = {"<value": [], "value<": []}
totSpam = 0
totNonSpam = 0
for line in file:
value = float(line[placeOfWordToCompare])
if value < valueToCompare:
if int(line[-1]) == 1:
spamClasses["<value"].append(value)
totSpam += 1
else:
nonSpamClasses["<value"].append(value)
totNonSpam += 1
else:
if int(line[-1]) == 1:
spamClasses["value<"].append(value)
totSpam += 1
else:
nonSpamClasses["value<"].append(value)
totNonSpam += 1
results = []
nbPlot = 1
for type in ["<value", "value<"]:
try:
totalType = len(spamClasses[type]) + len(nonSpamClasses[type])
totalSpam = len(spamClasses[type]) / totalType
totalNonSpam = len(nonSpamClasses[type]) / totalType
except:
totalType, totalSpam, totalNonSpam = 0, 0, 0
results.append(totalSpam)
results.append(totalNonSpam)
if draw:
labels = "Spam", "Non Spam"
sizes = [totalSpam, totalNonSpam]
colors = ['yellowgreen', 'lightskyblue']
plt.subplot(210 + nbPlot)
nbPlot += 1
plt.title("For : " + str(type) + " : " + str(valueToCompare) +
" for the word nb : " + str(placeOfWordToCompare) +
" with a number of " + str(totalType) +
" emails concerned")
plt.pie(sizes,
labels=labels,
colors=colors,
autopct='%1.1f%%',
shadow=True,
startangle=90)
if (results[0] * 100 > results[1] * 100 * precision
or results[1] * 100 > results[0] * 100 * precision):
if (results[2] * 100 > results[3] * 100 * precision
or results[3] * 100 > results[2] * 100 * precision):
#print("okk")
nbOk.append(placeOfWordToCompare)
if draw:
plt.axis('equal')
plt.savefig('PieChart01.png')
plt.show()
else:
if draw:
plt.close()
else:
if draw:
plt.close()
return nbOk
