from preprocess import splitColumns import numpy as np from NBCClassifier import featureVector from cvFunctions import getTrainData #from lr import lr from svm import svm #from nbc import nbc import math #from plot import plot # Defaults trainData = 'Data\yelp_data.csv' percentage = 100 # Get data as array train = getLines(trainData, 100) """ words is now a list of (word,frequency) tuples, ordered by descending order of frequency wordList is a list of all unique words in the training data """ """ The 100 most frequent words have been removed from words wordList is now a list of the words in desc order of frequency """ cv = [] #cv = [[[0 for z in range(4001)] for y in range(200)] for x in range(10)] #cv = np.array(cv);
def changeDepth(self, depths):
# Get data as array
train = getLines(self.trainData, 100)
cv = []
train = np.array(train)
np.random.shuffle(train)
for i in range(10):
cv.append(train[i * 200:(i + 1) * 200])
zoltempdt = [0 for xtemp in range(10)]
zoltemprf = [0 for xtemp in range(10)]
zoltempbag = [0 for xtemp in range(10)]
zoltempsvm = [0 for xtemp in range(10)]
w = 1000
it = 10
avgzoldt = [0 for xtemp in range(len(depths))]
avgzolrf = [0 for xtemp in range(len(depths))]
avgzolbag = [0 for xtemp in range(len(depths))]
avgzolsvm = [0 for xtemp in range(len(depths))]
stddevzoldt = [0 for xtemp in range(len(depths))]
stddevzolrf = [0 for xtemp in range(len(depths))]
stddevzolbag = [0 for xtemp in range(len(depths))]
stddevzolsvm = [0 for xtemp in range(len(depths))]
stderrzoldt = [0 for xtemp in range(len(depths))]
stderrzolrf = [0 for xtemp in range(len(depths))]
stderrzolbag = [0 for xtemp in range(len(depths))]
stderrzolsvm = [0 for xtemp in range(len(depths))]
testnew = []
trainnew = []
for r in range(len(depths)):
for i in range(it):
trainnew = []
testnew = cv[i]
for j in range(it):
if j != i:
for k in range(200):
trainnew.append(cv[j][k])
temptrain = trainnew
trainDataset = getTrainData(temptrain, 0.25)
rid_train, x_train, y_train = splitColumns(trainDataset)
rid_test, x_test, y_test = splitColumns(testnew)
# Pre-processing data
x_train = preprocess(x_train)
x_test = preprocess(x_test)
# Creating dictionary from x_train
words, wordList = getWordList(x_train)
# Removing most frequent 100 words
for _ in range(100):
words.pop(0)
wordList = [x for x, _ in words]
# Forming feature vector, calculating Conditional probabilities, applying bag
trainfv, trainfv0, trainfv1 = featureVector(
wordList[:w], x_train, y_train)
testfv, testfv0, testfv1 = featureVector(
wordList[:w], x_test, y_test)
zoltempdt[i] = decisionTree(trainfv, testfv, depths[r])
zoltemprf[i] = randomForest(trainfv, testfv, depths[r])
zoltempbag[i] = bagging(trainfv, testfv, depths[r])
# zoltempsvm[i] = svm(trainfv,testfv)
avgzoldt[r] = np.mean(zoltempdt)
avgzolrf[r] = np.mean(zoltemprf)
avgzolbag[r] = np.mean(zoltempbag)
avgzolsvm[r] = np.mean(zoltempsvm)
stddevzoldt[r] = np.std(zoltempdt)
stddevzolrf[r] = np.std(zoltemprf)
stddevzolbag[r] = np.std(zoltempbag)
stddevzolsvm[r] = np.std(zoltempsvm)
stderrzoldt[r] = stddevzoldt[r] / math.sqrt(it)
stderrzolrf[r] = stddevzolrf[r] / math.sqrt(it)
stderrzolbag[r] = stddevzolbag[r] / math.sqrt(it)
stderrzolsvm[r] = stddevzolsvm[r] / math.sqrt(it)
print avgzoldt
print avgzolbag
print avgzolrf
# print avgzolsvm
print stddevzoldt
print stddevzolbag
print stddevzolrf
# print stddevzolsvm
print stderrzoldt
print stderrzolbag
print stderrzolrf
# print stderrzolsvm
f = open(self.file, "a+")
f.write("\n AVERAGE ZERO ONE LOSS")
f.write("\n 1. Decision Tree")
f.write(str(avgzoldt))
f.write("\n 2. Bagging")
f.write(str(avgzolbag))
f.write("\n 3. Random forest")
f.write(str(avgzolrf))
# f.write("\n 4. SVM")
# f.write(str(avgzolsvm))
f.write("\n STANDARD DEVIATION ZERO ONE LOSS")
f.write("\n 1. Decision Tree")
f.write(str(stddevzoldt))
f.write("\n 2. Bagging")
f.write(str(stddevzolbag))
f.write("\n 3. Random forest")
f.write(str(stddevzolrf))
# f.write("\n 4. SVM")
# f.write(str(stddevzolsvm))
f.write("\n STANDARD ERROR ZERO ONE LOSS")
f.write("\n 1. Decision Tree")
f.write(str(stderrzoldt))
f.write("\n 2. Bagging")
f.write(str(stderrzolbag))
f.write("\n 3. Random forest")
f.write(str(stderrzolrf))
# f.write("\n 4. SVM")
# f.write(str(stderrzolsvm))
f.close()