from sklearn.linear_model import LogisticRegression

from sklearn.grid_search import GridSearchCV

import os

import numpy as np

import pickle

import json

from sklearn.metrics import accuracy_score

from sklearn.feature_extraction.text import CountVectorizer

from sklearn.naive_bayes import MultinomialNB, GaussianNB

from sklearn import svm

from sklearn.metrics import confusion_matrix

#Analysing the Age values in the dataset using Pandas' DataFrame

import pandas as pd

word2vec_path = '/Users/sunyambagga/Desktop/MinorProjects/7th_Sem/feature_vectors.pickle'

# Load the feature-vectors

with open(word2vec_path, 'rb') as f:

list_of_dicts = pickle.load(f)

df = pd.DataFrame(list_of_dicts)

df = df['Age']

print "\n\nPREDICTING AGE.."

print "\nAll bloggers in our dataset have ages b/w " + df.min() + " years and " + df.max() + " years."

###############################################################################

'''

1. First Approach:

- Using Average Word2Vec vectors for each blog (previously calculated and pickled in 'feature_extraction.py')

'''

print "\nFIRST APPROACH: Word2Vec.."

x = []

y = []

# Min Age- 13 & Max Age- 48

# Divide into Teens, 20s, 30s, 40s

#c13 = 0

#c20 = 0

#c30 = 0

#c40 = 0

for dict in list_of_dicts:

age = ''

if 13 <= int(dict['Age']) <= 19:

age = 'teens'

# c13 += 1

elif 20 <= int(dict['Age']) <= 29:

age = 'twenties'

# c20 += 1

elif 30 <= int(dict['Age']) <= 39:

age = 'thirties'

# c30 += 1

elif 40 <= int(dict['Age']) <= 49:

age = 'forties'

# c40 += 1

x.append(list(dict['BlogVector']))

y.append(age)

#print c13, c20, c30, c40

x_train = x[:16000]

y_train = y[:16000]

x_test = x[16000:]

y_test = y[16000:]

#print len(x)

#print len(y)

lr1_age_clf = LogisticRegression()

lr1_age_clf.fit(x_train, y_train)

blog = ""

print lr1_age_clf.predict([blog])

#y_pred = lr1_age_clf.predict(x_test)

from sklearn import metrics

print "\nLogistic Regression Accuracy: ", lr1_age_clf.score(x_test, y_test)

print "\nConfusion Matrix:\n", metrics.confusion_matrix(y_test, y_pred, labels=['teens', 'twenties', 'thirties', 'forties'])

print "\nClassification Report:\n", metrics.classification_report(y_test, y_pred, labels=['teens', 'twenties', 'thirties', 'forties'])

svm_age_clf = svm.LinearSVC()

svm_age_clf.fit(x_train, y_train)

print "\nSVM: ", svm_age_clf.score(x_test, y_test)

print "\n\n-------------------------------------------------------------------------------------------------------------------------"

##############################################################################

'''

2. Second Approach:

- Using Bag-of-Words feature-vectors for each blog (previously calculated and pickled in 'feature_extraction.py')

'''

print "\n\nSECOND APPROACH: Bag-of-Words.."

bow_path = '/Users/sunyambagga/Desktop/MinorProjects/7th_Sem/bow_features40.json'

# Load Important_Words

with open(bow_path, 'rb') as f:

_2_list_of_dicts = json.load(f)

#print "Loaded " + str(len(_2_list_of_dicts)) + " blogs."

# Prepare x_train/test and y_train/test

x_training = []

y_training = []

x_test = []

y_test = []

for dict in _2_list_of_dicts[:16000]:

words = dict['Imp_words']

sentence = " ".join(words)

x_training.append(sentence)

age = ''

if 13 <= int(dict['Age']) <= 19:

age = 'teens'

elif 20 <= int(dict['Age']) <= 29:

age = 'twenties'

elif 30 <= int(dict['Age']) <= 39:

age = 'thirties'

elif 40 <= int(dict['Age']) <= 49:

age = 'forties'

y_training.append(age)

for dict in _2_list_of_dicts[16000:]:

words = dict['Imp_words']

sentence = " ".join(words)

x_test.append(sentence)

age = ''

if 13 <= int(dict['Age']) <= 19:

age = 'teens'

elif 20 <= int(dict['Age']) <= 29:

age = 'twenties'

elif 30 <= int(dict['Age']) <= 39:

age = 'thirties'

elif 40 <= int(dict['Age']) <= 49:

age = 'forties'

y_test.append(age)

#print len(x_training)

#print len(y_training)

#print "\n\n"

#print len(x_test)

#print len(y_test)

vectorizer = CountVectorizer()

X = vectorizer.fit_transform(x_training)

X = X.toarray()

nb2_g_clf = MultinomialNB()

lr2_g_clf = LogisticRegression()

# Training Naive Bayes classifier for Gender:

nb2_g_clf.fit(X, y_training)

# Training LogReg classifier for Gender:

lr2_g_clf.fit(X, y_training)

X_test = vectorizer.transform(x_test)

print "\nNaive Bayes Accuracy: ", nb2_g_clf.score(X_test, y_test)

print "\nLogistic Regression Accuracy: ", lr2_g_clf.score(X_test, y_test)

svm_gender2_clf = svm.LinearSVC()

svm_gender2_clf.fit(X, y_training)

print "\nSVM Accuracy: ", svm_gender2_clf.score(X_test, y_test)

print "\n\n\n\n"

##############################################################################