#!/usr/bin/env python
# coding: utf-8

# In[3]:


import pathlib
from collections import Counter
from sklearn.metrics import *
import pandas as pd


# In[1]:


import numpy as np, pandas as pd
import seaborn as sns
import matplotlib.pyplot as plt
from sklearn.datasets import fetch_20newsgroups
from sklearn.feature_extraction.text import TfidfVectorizer
from sklearn.naive_bayes import MultinomialNB
from sklearn.pipeline import make_pipeline
from sklearn.metrics import confusion_matrix, accuracy_score
sns.set() # use seaborn plotting style


# In[5]:


train_x = pd.read_csv('train/in.tsv', header=None, sep='\t')
train_y = pd.read_csv('train/expected.tsv', header=None, sep='\t')
dev_x = pd.read_csv('dev-0/in.tsv', header=None, sep='\t')
dev_y = pd.read_csv('dev-0/expected.tsv', header=None, sep='\t')
test_x = pd.read_csv('test-A/in.tsv', header=None, sep='\t')


# In[61]:


print(dev_y.shape)
print(dev_x.shape)


# In[11]:


print(train_x[:15])


# In[27]:


print(train_x.shape)


# In[49]:


print(train_y.shape)


# In[8]:


print(train_y[:15])


# In[53]:


print(dev_x[:4])


# In[119]:


from sklearn.feature_extraction.text import CountVectorizer
from sklearn.feature_extraction.text import TfidfTransformer

vec = CountVectorizer(stop_words='english')
x1 = vec.fit_transform(train_x[:20000][0])
tfidf_transformer = TfidfTransformer()

x1_tf = tfidf_transformer.fit_transform(x1)


# In[120]:


# Build the model
#model = make_pipeline(TfidfVectorizer(), MultinomialNB())
clf = MultinomialNB().fit(x1_tf, train_y[:20000][0])


# In[121]:


# Train the model using the training data
#model.fit(x1[:][0], train_y[:289541][0])
# Predict the categories of the test data
X_new_counts = vec.transform(dev_x[:][0])
# We call transform instead of fit_transform because it's already been fit
X_new_tfidf = tfidf_transformer.transform(X_new_counts)
#predicted_categories = model.predict(dev_x[:][0])


# In[122]:


predicted = clf.predict(X_new_tfidf)


# In[125]:


print(predicted[:10])


# In[126]:


print(predicted.shape)


# In[123]:


#mat = confusion_matrix(dev_y[:][0],predicted_categories)

print("The accuracy is {}".format(accuracy_score( dev_y[:][0],predicted_categories)))
 


# In[124]:


print('We got an accuracy of',np.mean(predicted == dev_y[:][0])*100, '% over the test data.')


# In[130]:


np.savetxt("out.tsv",predicted, delimiter="\t", fmt='%d')


# In[131]:


X_test = vec.transform(test_x[:][0])
# We call transform instead of fit_transform because it's already been fit
X_tfidf_test = tfidf_transformer.transform(X_test)
predicted_test = clf.predict(X_tfidf_test)
np.savetxt("out.tsv",predicted_test, delimiter="\t", fmt='%d')


# In[ ]: