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

# In[1]:


import os
import pandas as pd
import numpy as np
import sklearn
from sklearn.feature_extraction.text import TfidfVectorizer
from sklearn.linear_model import LinearRegression
from sklearn.metrics import mean_squared_error
from sklearn.pipeline import make_pipeline


# In[2]:


train = pd.read_csv('train/train.tsv', header=None, sep='\t', error_bad_lines=False)
print(len(train))
train = train.head(40000)


# In[3]:


x_train = train[4]
y_train = train[0]


# In[4]:


x_dev_data = pd.read_csv('dev-0/in.tsv', header=None, sep='\t')
x_dev = x_dev_data[0]
x_dev[19999] = "to jest tekst testowy"
x_dev[20000] = "a ten tekst jest najbardziej testowy"
y_dev = pd.read_csv('dev-0/expected.tsv', header=None, sep='\t')


# In[5]:


model = make_pipeline(TfidfVectorizer(), LinearRegression())
model.fit(x_train, y_train)


# In[6]:


dev_predicted =  model.predict(x_dev)

with open('dev-0/out.tsv', 'wt') as f:
    for i in dev_predicted:
        f.write(str(i)+'\n')

dev_out = pd.read_csv('dev-0/out.tsv', header=None, sep='\t')
dev_expected = pd.read_csv('dev-0/expected.tsv', header=None, sep='\t')


# In[7]:


print(mean_squared_error(dev_out, dev_expected))


# In[8]:


with open('test-A/in.tsv', 'r', encoding = 'utf-8') as f:
       x_test = f.readlines()
        
# x_test = pd.Series(x_test)
# x_test = vectorizer.transform(x_test)

test_predicted = model.predict(x_test)

with open('test-A/out.tsv', 'wt') as f:
    for i in test_predicted:
        f.write(str(i)+'\n')


# In[9]:


get_ipython().system('jupyter nbconvert --to script run.ipynb')