import numpy as np 
import pandas as pd
import matplotlib.pyplot as plt
import datetime

train = pd.read_csv('train_file.csv')
test =  pd.read_csv('test_file.csv')
submission = pd.read_csv('sample_submission.csv')

test_id = test['IDLink']

Data Preprocessing

!pip3 install afinn

Requirement already satisfied: afinn in /usr/local/lib/python3.6/dist-packages (0.1)

from textblob import TextBlob
import nltk
nltk.download('stopwords')
nltk.download('punkt')
nltk.download('wordnet')
import re
from nltk.tokenize import word_tokenize
from nltk.corpus import stopwords
stop = set(stopwords.words('english'))

[nltk_data] Downloading package stopwords to /root/nltk_data...
[nltk_data]   Package stopwords is already up-to-date!
[nltk_data] Downloading package punkt to /root/nltk_data...
[nltk_data]   Package punkt is already up-to-date!
[nltk_data] Downloading package wordnet to /root/nltk_data...
[nltk_data]   Package wordnet is already up-to-date!

from textblob import Word 

from sklearn.feature_extraction.text import TfidfVectorizer
from sklearn.preprocessing import LabelEncoder, StandardScaler,MinMaxScaler
from sklearn.model_selection import train_test_split
from scipy.sparse import hstack, csr_matrix

train.head()

	IDLink	Title	Headline	Source	Topic	PublishDate	Facebook	GooglePlus	LinkedIn	SentimentTitle	SentimentHeadline
0	Tr3CMgRv1N	Obama Lays Wreath at Arlington National Cemetery	Obama Lays Wreath at Arlington National Cemete...	USA TODAY	obama	2002-04-02 00:00:00	-1	-1	-1	0.000000	-0.053300
1	Wc81vGp8qZ	A Look at the Health of the Chinese Economy	Tim Haywood, investment director business-unit...	Bloomberg	economy	2008-09-20 00:00:00	-1	-1	-1	0.208333	-0.156386
2	zNGH03CrZH	Nouriel Roubini: Global Economy Not Back to 2008	Nouriel Roubini, NYU professor and chairman at...	Bloomberg	economy	2012-01-28 00:00:00	-1	-1	-1	-0.425210	0.139754
3	3sM1H0W8ts	Finland GDP Expands In Q4	Finland's economy expanded marginally in the t...	RTT News	economy	2015-03-01 00:06:00	-1	-1	-1	0.000000	0.026064
4	wUbnxgvqaZ	Tourism, govt spending buoys Thai economy in J...	Tourism and public spending continued to boost...	The Nation - Thailand's English news	economy	2015-03-01 00:11:00	-1	-1	-1	0.000000	0.141084

train.describe()

	Facebook	GooglePlus	LinkedIn	SentimentTitle	SentimentHeadline
count	55932.000000	55932.000000	55932.000000	55932.000000	55932.000000
mean	132.050329	4.551616	14.300132	-0.006318	-0.029577
std	722.931314	21.137177	76.651420	0.137569	0.143038
min	-1.000000	-1.000000	-1.000000	-0.838525	-0.755355
25%	0.000000	0.000000	0.000000	-0.079057	-0.116927
50%	6.000000	0.000000	0.000000	0.000000	-0.027277
75%	37.000000	2.000000	4.000000	0.063969	0.057354
max	49211.000000	1267.000000	3716.000000	0.962354	0.964646

test.describe()

	Facebook	GooglePlus	LinkedIn
count	37288.000000	37288.000000	37288.000000
mean	84.832654	2.895516	19.928181
std	420.063477	13.538942	225.443395
min	-1.000000	-1.000000	-1.000000
25%	0.000000	0.000000	0.000000
50%	4.000000	0.000000	0.000000
75%	28.000000	2.000000	4.000000
max	16598.000000	1016.000000	20341.000000

train.isna().sum()

IDLink                 0
Title                  0
Headline               0
Source               175
Topic                  0
PublishDate            0
Facebook               0
GooglePlus             0
LinkedIn               0
SentimentTitle         0
SentimentHeadline      0
dtype: int64

test.isna().sum()

IDLink           0
Title            0
Headline         0
Source         101
Topic            0
PublishDate      0
Facebook         0
GooglePlus       0
LinkedIn         0
dtype: int64

train['Source'].mode()

0    Bloomberg
dtype: object

train['Source'] = train['Source'].fillna('Bloomberg')
test['Source'] = test['Source'].fillna('Bloomberg')

Clean the Text : Remove unwanted characters and stop words

train['Text_Title'] = train['Title'] + ' ' + train['Source'] + ' ' + train['Topic']
test['Text_Title'] = test['Title'] + ' ' + test['Source'] + ' ' + test['Topic']

train['Text_Headline'] = train['Headline'] + ' ' + train['Source'] + ' ' + train['Topic'] 
test['Text_Headline'] = test['Headline'] + ' ' + test['Source'] + ' ' + test['Topic'] 

def clean_data(text):
  text_token = word_tokenize(text)
  filtered_text = ' '.join([w.lower() for w in text_token if w.lower() not in stop and len(w) > 2]) #Converting to lower cases and removing stop words
  filtered_text = filtered_text.replace(r"[^a-zA-Z]+", '') #remove special characters
  text_only = re.sub(r'\b\d+\b', '', filtered_text)   #remove numbers
  clean_text = text_only.replace(',', '').replace('.', '').replace(':', '')
  return clean_text

def lemmatize(text):
  text_token = word_tokenize(text)
  lematized_text = ' '.join([Word(w).lemmatize() for w in text_token]) # Lematization
  return lematized_text

train['Text_Title'][2]

'Nouriel Roubini: Global Economy Not Back to 2008 Bloomberg economy'

train['Text_Title'] = [clean_data(x) for x in train['Text_Title']]
test['Text_Title'] = [clean_data(x) for x in test['Text_Title']]

train['Text_Headline'] = [clean_data(x) for x in train['Text_Headline']]
test['Text_Headline'] = [clean_data(x) for x in test['Text_Headline']]

train['Text_Title'][2]

'nouriel roubini global economy back  bloomberg economy'

# train['Text_Title'] = [lemmatize(x) for x in train['Text_Title']]
# test['Text_Title'] = [lemmatize(x) for x in test['Text_Title']]

# train['Text_Headline'] = [lemmatize(x) for x in train['Text_Headline']]
# test['Text_Headline'] = [lemmatize(x) for x in test['Text_Headline']]

# train['Text_Title'][2]

Vectorizing text

vectorizer = TfidfVectorizer(use_idf=True)

train_v_Title = vectorizer.fit_transform(train['Text_Title'])
test_v_Title = vectorizer.transform(test['Text_Title'])

vectorizer_ = TfidfVectorizer(use_idf=True)

train_v_Headline = vectorizer_.fit_transform(train['Text_Headline'])
test_v_Headline = vectorizer_.transform(test['Text_Headline'])

test_v_Headline.shape

(37288, 38787)

Sentiment Extraction with affin and textblob polarity and subjectivity

from afinn import Afinn
af = Afinn()

Title

train['pol_t'] = train['Title'].apply(lambda x: TextBlob(x).sentiment.polarity)
test['pol_t'] = test['Title'].apply(lambda x: TextBlob(x).sentiment.polarity)

train['sub_t'] = train['Title'].apply(lambda x: TextBlob(x).sentiment.subjectivity)
test['sub_t'] = test['Title'].apply(lambda x: TextBlob(x).sentiment.subjectivity)

train['aff_t'] = train['Title'].apply(lambda x: af.score(x))
test['aff_t'] = test['Title'].apply(lambda x: af.score(x))

Headline

train['pol_h'] = train['Headline'].apply(lambda x: TextBlob(x).sentiment.polarity)
test['pol_h'] = test['Headline'].apply(lambda x: TextBlob(x).sentiment.polarity)

train['sub_h'] = train['Headline'].apply(lambda x: TextBlob(x).sentiment.subjectivity)
test['sub_h'] = test['Headline'].apply(lambda x: TextBlob(x).sentiment.subjectivity)

train['aff_h'] = train['Headline'].apply(lambda x: af.score(x))
test['aff_h'] = test['Headline'].apply(lambda x: af.score(x))

Feature Engineering

Label Encode categorical Variables

encoder = LabelEncoder()

train['Topic'] = encoder.fit_transform(train['Topic'])
test['Topic'] = encoder.transform(test['Topic'])

total = train['Source'].to_list() + test['Source'].to_list()
total = encoder.fit_transform(total)
train['Source'] = encoder.transform(train['Source'])
test['Source'] = encoder.transform(test['Source'])

Processing the date time stamp

train["month_start"] = train["PublishDate"].apply(lambda x: 1 if int(x.split()[0].split(':')[0].split('-')[2])<= 7 else 0)
test["month_start"] = test["PublishDate"].apply(lambda x: 1 if int(x.split()[0].split(':')[0].split('-')[2])<= 7 else 0)

train["month_end"] = train["PublishDate"].apply(lambda x: 1 if int(x.split()[0].split(':')[0].split('-')[2])>= 24 else 0)
test["month_end"] = test["PublishDate"].apply(lambda x: 1 if int(x.split()[0].split(':')[0].split('-')[2]) >=24 else 0)

train["month_mid"] = train["PublishDate"].apply(lambda x: 1 if (int(x.split()[0].split(':')[0].split('-')[2]) < 24) and (int(x.split()[0].split(':')[0].split('-')[2]) > 7) else 0)
test["month_mid"] = test["PublishDate"].apply(lambda x: 1 if (int(x.split()[0].split(':')[0].split('-')[2]) < 24) and (int(x.split()[0].split(':')[0].split('-')[2]) > 7) else 0)

# Hour from date
train["hour"] = train["PublishDate"].apply(lambda x: x.split()[1].split(':')[0])
test["hour"] = test["PublishDate"].apply(lambda x: x.split()[1].split(':')[0])

#year from the date
train["year"] = train["PublishDate"].apply(lambda x: x.split()[0].split(':')[0].split('-')[0])
test["year"] = test["PublishDate"].apply(lambda x: x.split()[0].split(':')[0].split('-')[0])

#Month from the date
train["month"] = train["PublishDate"].apply(lambda x: x.split()[0].split(':')[0].split('-')[1])
test["month"] = test["PublishDate"].apply(lambda x: x.split()[0].split(':')[0].split('-')[1])

Plots

# Month distribution of SentimentTitle
plt.scatter(train['month'], train['SentimentTitle'],c='y')
plt.xlabel('month')
plt.ylabel('SentimentTitle')

Text(0, 0.5, 'SentimentTitle')

# Month distribution of SentimentTitle
plt.scatter(train['month'], train['SentimentHeadline'] ,c='y')
plt.xlabel('month')
plt.ylabel('SentimentHeadline')

Text(0, 0.5, 'SentimentHeadline')

# hour distribution of SentimentTitle

plt.scatter(train['hour'], train['SentimentTitle'],c='g')
plt.xlabel('hour')
plt.ylabel('SentimentTitle')

Text(0, 0.5, 'SentimentTitle')

# hour distribution of SentimentHeadline
plt.scatter(train['hour'], train['SentimentHeadline'],c= 'g')
plt.xlabel('hour')
plt.ylabel('SentimentHeadline')

Text(0, 0.5, 'SentimentHeadline')

#month start = yellow
plt.scatter(train['month'],train['SentimentHeadline'],c = train.month_start)
plt.xlabel('month')
plt.ylabel('sentiment Headline')

Text(0, 0.5, 'sentiment Headline')

#month mid = yellow
plt.scatter(train['month'],train['SentimentHeadline'],c = train.month_mid)
plt.xlabel('month')
plt.ylabel('sentiment Headline')

Text(0, 0.5, 'sentiment Headline')

#month end = yellow
plt.scatter(train['month'],train['SentimentHeadline'],c = train.month_end)
plt.xlabel('month')
plt.ylabel('sentiment Headline')

Text(0, 0.5, 'sentiment Headline')

Descriptive Variables

Standardising

scaler = StandardScaler()

cols = ['Source', 'Topic', 'Facebook', 'GooglePlus', 'LinkedIn','hour', 'year','month']

for col in cols:
  train[col] = scaler.fit_transform(train[col].values.reshape(-1, 1))
  test[col] = scaler.transform(test[col].values.reshape(-1, 1))

cols_t = ['Source', 'Topic', 'Facebook', 'LinkedIn', 'GooglePlus','pol_t','aff_t', 'sub_t', 'hour','year','month','month_start' ,'month_end','month_mid']
train_X1 = train[cols_t]
test_X1 = test[cols_t]

cols_h = ['Source', 'Topic', 'Facebook', 'GooglePlus', 'LinkedIn','aff_h','pol_h', 'sub_h', 'hour','year','month_start' ,'month_mid','month']
train_X2 = train[cols_h]
test_X2 = test[cols_h]

d= train_X2.copy()
d['target'] = train['SentimentHeadline']

import seaborn as sn
plt.figure(figsize = (16,5))
sn.heatmap(d.corr(), annot=True)

plt.show()

train_X2.head()

	Source	Topic	Facebook	GooglePlus	LinkedIn	aff_h	aff_t	pol_h	sub_h	hour	year	month_start	month_mid	month
0	1.257700	0.841443	-0.184044	-0.262649	-0.199608	2.0	0.0	0.000000	0.000000	-1.699073	-27.994016	1	0	-0.312446
1	-1.314091	-1.108773	-0.184044	-0.262649	-0.199608	0.0	0.0	0.100000	0.200000	-1.699073	-15.672447	0	1	0.753237
2	-1.314091	-1.108773	-0.184044	-0.262649	-0.199608	0.0	0.0	0.000000	0.041667	-1.699073	-7.458068	0	0	-0.951856
3	0.468412	-1.108773	-0.184044	-0.262649	-0.199608	0.0	1.0	-0.166667	0.166667	-1.699073	-1.297284	1	0	-0.525582
4	1.029071	-1.108773	-0.184044	-0.262649	-0.199608	1.0	0.0	0.133333	0.380556	-1.699073	-1.297284	1	0	-0.525582

print(np.shape(train_X1))
print(np.shape(test_X1))
print(np.shape(train_X2))
print(np.shape(test_X2))
print(np.shape(train_v_Title))
print(np.shape(test_v_Title))
print(np.shape(train_v_Headline))
print(np.shape(test_v_Headline))

(55932, 15)
(37288, 15)
(55932, 14)
(37288, 14)
(55932, 25377)
(37288, 25377)
(55932, 38787)
(37288, 38787)

type(train_v_Title)

scipy.sparse.csr.csr_matrix

train_X_Title = hstack([train_v_Title, csr_matrix(train_X1.values)]) #merging vectorised words and additional variables
test_X_Title = hstack([test_v_Title, csr_matrix(test_X1.values)])
y1 = train['SentimentTitle']

train_X_Headline = hstack([train_v_Headline, csr_matrix(train_X2.values)])
test_X_Headline = hstack([test_v_Headline, csr_matrix(test_X2.values)])
y2 = train['SentimentHeadline']

NN

from sklearn.neural_network import MLPRegressor

mlp = MLPRegressor(hidden_layer_sizes=50,activation='tanh',batch_size = 160,alpha = 0.05,verbose = True,random_state = 1,warm_start = True,tol = 0.000001)

X_train, X_test, y_train, y_test = train_test_split(train_X_Headline, y2, test_size=0.2, random_state=42)

mlp.alpha = 0.2

mlp.batch_size = 412

mlp.activation= 'tanh'

mlp.fit(X_train, y_train)

Iteration 63, loss = 0.00743567
Iteration 64, loss = 0.00726868
Iteration 65, loss = 0.00723798
Iteration 66, loss = 0.00722085
Iteration 67, loss = 0.00723858
Iteration 68, loss = 0.00723238

/usr/local/lib/python3.6/dist-packages/sklearn/neural_network/_multilayer_perceptron.py:573: UserWarning: Training interrupted by user.
  warnings.warn("Training interrupted by user.")

MLPRegressor(activation='tanh', alpha=0.2, batch_size=412, beta_1=0.9,
             beta_2=0.999, early_stopping=False, epsilon=1e-08,
             hidden_layer_sizes=50, learning_rate='constant',
             learning_rate_init=0.001, max_fun=15000, max_iter=200,
             momentum=0.9, n_iter_no_change=10, nesterovs_momentum=True,
             power_t=0.5, random_state=1, shuffle=True, solver='adam',
             tol=1e-06, validation_fraction=0.1, verbose=True, warm_start=True)

y_pred1 = mlp.predict(X_test)
y_pred2 = mlp.predict(X_train)
mae1 = mean_absolute_error(y_pred1, y_test)
mae2 = mean_absolute_error(y_pred2, y_train)
print('MAE test:', 1 - mae1)
print('MAE train:', 1 - mae2)

MAE test: 0.9222385602688256
MAE train: 0.9299610546514606

mlp1 = MLPRegressor(hidden_layer_sizes=20,activation='tanh',batch_size = 64,alpha = 0.02,verbose = True,random_state = 2,warm_start = True,tol = 0.000001)

X_train, X_test, y_train, y_test = train_test_split(train_X_Headline, y2, test_size=0.20, random_state=42)
mlp1.fit(X_train, y_train)
y_pred1 = mlp1.predict(X_train)
mae2 = mean_absolute_error(y_pred1, y_train)
print('MAE:', 1 - mae2)

MAE: 0.9045360123539832

/usr/local/lib/python3.6/dist-packages/sklearn/neural_network/_multilayer_perceptron.py:573: UserWarning: Training interrupted by user.
  warnings.warn("Training interrupted by user.")

Ensemble

X_train, X_test, y_train, y_test = train_test_split(train_X_Headline, y2, test_size=0.25,random_state = 32)

import xgboost as xgb
dtrain = xgb.DMatrix(X_train, label=y_train)
dtest = xgb.DMatrix(X_test, label=y_test)

params = {
    # Parameters that we are going to tune.
    'max_depth':6,
    'min_child_weight': 1,
    'eta':.3,
    'subsample': 1,
    'colsample_bytree': 1,
    # Other parameters
    'objective':'reg:linear',
    'eval_metric':"mae"
}

model = xgb.train(
    params,
    dtrain,
    num_boost_round=999,
    evals=[(dtest, "Test")],
    early_stopping_rounds=10
)

[06:28:41] WARNING: /workspace/src/objective/regression_obj.cu:152: reg:linear is now deprecated in favor of reg:squarederror.
[0]	Test-mae:0.370408
Will train until Test-mae hasn't improved in 10 rounds.
[1]	Test-mae:0.262108
[2]	Test-mae:0.191631
[3]	Test-mae:0.149143
[4]	Test-mae:0.12576
[5]	Test-mae:0.112916
[6]	Test-mae:0.106259
[7]	Test-mae:0.102671
[8]	Test-mae:0.100707
[9]	Test-mae:0.099683
[10]	Test-mae:0.098989
[11]	Test-mae:0.09839
[12]	Test-mae:0.097973
[13]	Test-mae:0.097607
[14]	Test-mae:0.097271
[15]	Test-mae:0.096879
[16]	Test-mae:0.096592
[17]	Test-mae:0.096431
[18]	Test-mae:0.096205
[19]	Test-mae:0.095895
[20]	Test-mae:0.095611
[21]	Test-mae:0.095384
[22]	Test-mae:0.094964
[23]	Test-mae:0.094715
[24]	Test-mae:0.094457
[25]	Test-mae:0.094258
[26]	Test-mae:0.093987
[27]	Test-mae:0.093809
[28]	Test-mae:0.09365
[29]	Test-mae:0.093506
[30]	Test-mae:0.093317
[31]	Test-mae:0.093151
[32]	Test-mae:0.092999
[33]	Test-mae:0.092812
[34]	Test-mae:0.092635
[35]	Test-mae:0.092445
[36]	Test-mae:0.092185
[37]	Test-mae:0.092027
[38]	Test-mae:0.091871
[39]	Test-mae:0.091733
[40]	Test-mae:0.091592
[41]	Test-mae:0.091489
[42]	Test-mae:0.091344
[43]	Test-mae:0.09124
[44]	Test-mae:0.091143
[45]	Test-mae:0.091028
[46]	Test-mae:0.090931
[47]	Test-mae:0.090784
[48]	Test-mae:0.090693
[49]	Test-mae:0.090544
[50]	Test-mae:0.090443
[51]	Test-mae:0.090305
[52]	Test-mae:0.090149
[53]	Test-mae:0.09009
[54]	Test-mae:0.08993
[55]	Test-mae:0.089802
[56]	Test-mae:0.089677
[57]	Test-mae:0.089556
[58]	Test-mae:0.089466
[59]	Test-mae:0.089378
[60]	Test-mae:0.089278
[61]	Test-mae:0.089169
[62]	Test-mae:0.089084
[63]	Test-mae:0.088985
[64]	Test-mae:0.088859
[65]	Test-mae:0.088733
[66]	Test-mae:0.088582
[67]	Test-mae:0.088486
[68]	Test-mae:0.088394
[69]	Test-mae:0.08832
[70]	Test-mae:0.088254
[71]	Test-mae:0.088152
[72]	Test-mae:0.08807
[73]	Test-mae:0.087976
[74]	Test-mae:0.087879
[75]	Test-mae:0.087825
[76]	Test-mae:0.087767
[77]	Test-mae:0.087724
[78]	Test-mae:0.087657
[79]	Test-mae:0.087549
[80]	Test-mae:0.087466
[81]	Test-mae:0.087387
[82]	Test-mae:0.087312
[83]	Test-mae:0.087196
[84]	Test-mae:0.087098
[85]	Test-mae:0.087034
[86]	Test-mae:0.086903
[87]	Test-mae:0.086867
[88]	Test-mae:0.086833
[89]	Test-mae:0.086708
[90]	Test-mae:0.086661
[91]	Test-mae:0.086581
[92]	Test-mae:0.086455
[93]	Test-mae:0.086355
[94]	Test-mae:0.086272
[95]	Test-mae:0.08618
[96]	Test-mae:0.086091
[97]	Test-mae:0.08602
[98]	Test-mae:0.085969
[99]	Test-mae:0.085933
[100]	Test-mae:0.085846
[101]	Test-mae:0.085814
[102]	Test-mae:0.085742
[103]	Test-mae:0.085702
[104]	Test-mae:0.085679
[105]	Test-mae:0.085645
[106]	Test-mae:0.085595
[107]	Test-mae:0.08553
[108]	Test-mae:0.085434
[109]	Test-mae:0.085382
[110]	Test-mae:0.085306
[111]	Test-mae:0.085292
[112]	Test-mae:0.085234
[113]	Test-mae:0.08521
[114]	Test-mae:0.085163
[115]	Test-mae:0.085036
[116]	Test-mae:0.085001
[117]	Test-mae:0.084933
[118]	Test-mae:0.084899
[119]	Test-mae:0.084845
[120]	Test-mae:0.084806
[121]	Test-mae:0.084755
[122]	Test-mae:0.084689
[123]	Test-mae:0.08464
[124]	Test-mae:0.08462
[125]	Test-mae:0.084554
[126]	Test-mae:0.084547
[127]	Test-mae:0.084506
[128]	Test-mae:0.084468
[129]	Test-mae:0.084458
[130]	Test-mae:0.084429
[131]	Test-mae:0.084353
[132]	Test-mae:0.084295
[133]	Test-mae:0.084248
[134]	Test-mae:0.084206
[135]	Test-mae:0.084169
[136]	Test-mae:0.084143
[137]	Test-mae:0.084116
[138]	Test-mae:0.084054
[139]	Test-mae:0.083992
[140]	Test-mae:0.083971
[141]	Test-mae:0.083915
[142]	Test-mae:0.083849
[143]	Test-mae:0.083797
[144]	Test-mae:0.08376
[145]	Test-mae:0.083721
[146]	Test-mae:0.083678
[147]	Test-mae:0.083646
[148]	Test-mae:0.083581
[149]	Test-mae:0.083522
[150]	Test-mae:0.083502
[151]	Test-mae:0.083463
[152]	Test-mae:0.083436
[153]	Test-mae:0.083398
[154]	Test-mae:0.083366
[155]	Test-mae:0.083308
[156]	Test-mae:0.083286
[157]	Test-mae:0.083232
[158]	Test-mae:0.083219
[159]	Test-mae:0.083199
[160]	Test-mae:0.083171
[161]	Test-mae:0.083133
[162]	Test-mae:0.083103
[163]	Test-mae:0.08308
[164]	Test-mae:0.083048
[165]	Test-mae:0.083023
[166]	Test-mae:0.082971
[167]	Test-mae:0.08293
[168]	Test-mae:0.082922
[169]	Test-mae:0.082874
[170]	Test-mae:0.082862
[171]	Test-mae:0.082844
[172]	Test-mae:0.082793
[173]	Test-mae:0.082743
[174]	Test-mae:0.08272
[175]	Test-mae:0.082683
[176]	Test-mae:0.082652
[177]	Test-mae:0.082613
[178]	Test-mae:0.082586
[179]	Test-mae:0.082564
[180]	Test-mae:0.082543
[181]	Test-mae:0.082548
[182]	Test-mae:0.082525
[183]	Test-mae:0.082477
[184]	Test-mae:0.082421
[185]	Test-mae:0.082357
[186]	Test-mae:0.082369
[187]	Test-mae:0.082323
[188]	Test-mae:0.082306
[189]	Test-mae:0.082279
[190]	Test-mae:0.08225
[191]	Test-mae:0.082223
[192]	Test-mae:0.082207
[193]	Test-mae:0.082175
[194]	Test-mae:0.082137
[195]	Test-mae:0.082113
[196]	Test-mae:0.082118
[197]	Test-mae:0.08209
[198]	Test-mae:0.082027
[199]	Test-mae:0.082007
[200]	Test-mae:0.081973
[201]	Test-mae:0.081981
[202]	Test-mae:0.081937
[203]	Test-mae:0.081899
[204]	Test-mae:0.081878
[205]	Test-mae:0.081857
[206]	Test-mae:0.081854
[207]	Test-mae:0.081813
[208]	Test-mae:0.081785
[209]	Test-mae:0.081748
[210]	Test-mae:0.08173
[211]	Test-mae:0.081718
[212]	Test-mae:0.081703
[213]	Test-mae:0.081671
[214]	Test-mae:0.081633
[215]	Test-mae:0.081607
[216]	Test-mae:0.081595
[217]	Test-mae:0.081575
[218]	Test-mae:0.081539
[219]	Test-mae:0.081502
[220]	Test-mae:0.081475
[221]	Test-mae:0.081462
[222]	Test-mae:0.081417
[223]	Test-mae:0.08139
[224]	Test-mae:0.081351
[225]	Test-mae:0.081349
[226]	Test-mae:0.081312
[227]	Test-mae:0.081281
[228]	Test-mae:0.081258
[229]	Test-mae:0.081257
[230]	Test-mae:0.081247
[231]	Test-mae:0.081216
[232]	Test-mae:0.081214
[233]	Test-mae:0.081209
[234]	Test-mae:0.081207
[235]	Test-mae:0.081176
[236]	Test-mae:0.081165
[237]	Test-mae:0.081156
[238]	Test-mae:0.081157
[239]	Test-mae:0.081119
[240]	Test-mae:0.0811
[241]	Test-mae:0.081069
[242]	Test-mae:0.081048
[243]	Test-mae:0.08105
[244]	Test-mae:0.081036
[245]	Test-mae:0.081014
[246]	Test-mae:0.080983
[247]	Test-mae:0.080951
[248]	Test-mae:0.0809
[249]	Test-mae:0.08087
[250]	Test-mae:0.080853
[251]	Test-mae:0.080825
[252]	Test-mae:0.080803
[253]	Test-mae:0.080779
[254]	Test-mae:0.08076
[255]	Test-mae:0.080759
[256]	Test-mae:0.080755
[257]	Test-mae:0.080731
[258]	Test-mae:0.080707
[259]	Test-mae:0.080673
[260]	Test-mae:0.080637
[261]	Test-mae:0.080607
[262]	Test-mae:0.080594
[263]	Test-mae:0.080581
[264]	Test-mae:0.080573
[265]	Test-mae:0.080556
[266]	Test-mae:0.080544
[267]	Test-mae:0.080539
[268]	Test-mae:0.080515
[269]	Test-mae:0.080494
[270]	Test-mae:0.080466
[271]	Test-mae:0.080436
[272]	Test-mae:0.080417
[273]	Test-mae:0.080403
[274]	Test-mae:0.080401
[275]	Test-mae:0.080382
[276]	Test-mae:0.080351
[277]	Test-mae:0.080325
[278]	Test-mae:0.08031
[279]	Test-mae:0.080307
[280]	Test-mae:0.080283
[281]	Test-mae:0.080262
[282]	Test-mae:0.080248
[283]	Test-mae:0.080225
[284]	Test-mae:0.080192
[285]	Test-mae:0.080172
[286]	Test-mae:0.080152
[287]	Test-mae:0.080149
[288]	Test-mae:0.080123
[289]	Test-mae:0.080104
[290]	Test-mae:0.0801
[291]	Test-mae:0.080082
[292]	Test-mae:0.080078
[293]	Test-mae:0.080068
[294]	Test-mae:0.080037
[295]	Test-mae:0.08002
[296]	Test-mae:0.080006
[297]	Test-mae:0.079989
[298]	Test-mae:0.079968
[299]	Test-mae:0.079942
[300]	Test-mae:0.079895
[301]	Test-mae:0.079867
[302]	Test-mae:0.079856
[303]	Test-mae:0.079856
[304]	Test-mae:0.07983
[305]	Test-mae:0.079806
[306]	Test-mae:0.079791
[307]	Test-mae:0.079781
[308]	Test-mae:0.079758
[309]	Test-mae:0.079753
[310]	Test-mae:0.079729
[311]	Test-mae:0.079727
[312]	Test-mae:0.07973
[313]	Test-mae:0.079703
[314]	Test-mae:0.079692
[315]	Test-mae:0.079686
[316]	Test-mae:0.079663
[317]	Test-mae:0.079657
[318]	Test-mae:0.079644
[319]	Test-mae:0.079631
[320]	Test-mae:0.079618
[321]	Test-mae:0.0796
[322]	Test-mae:0.079596
[323]	Test-mae:0.079581
[324]	Test-mae:0.079559
[325]	Test-mae:0.079534
[326]	Test-mae:0.079521
[327]	Test-mae:0.079502
[328]	Test-mae:0.079486
[329]	Test-mae:0.079474
[330]	Test-mae:0.079471
[331]	Test-mae:0.079468
[332]	Test-mae:0.07946
[333]	Test-mae:0.079453
[334]	Test-mae:0.079416
[335]	Test-mae:0.079402
[336]	Test-mae:0.079392
[337]	Test-mae:0.079384
[338]	Test-mae:0.079374
[339]	Test-mae:0.079382
[340]	Test-mae:0.079368
[341]	Test-mae:0.079365
[342]	Test-mae:0.079342
[343]	Test-mae:0.079331
[344]	Test-mae:0.07933
[345]	Test-mae:0.079314
[346]	Test-mae:0.079308
[347]	Test-mae:0.079289
[348]	Test-mae:0.079274
[349]	Test-mae:0.079266
[350]	Test-mae:0.079243
[351]	Test-mae:0.079207
[352]	Test-mae:0.079189
[353]	Test-mae:0.079165
[354]	Test-mae:0.079145
[355]	Test-mae:0.079118
[356]	Test-mae:0.079134
[357]	Test-mae:0.079117
[358]	Test-mae:0.079101
[359]	Test-mae:0.079102
[360]	Test-mae:0.079084
[361]	Test-mae:0.079085
[362]	Test-mae:0.07908
[363]	Test-mae:0.079067
[364]	Test-mae:0.079053
[365]	Test-mae:0.079036
[366]	Test-mae:0.079015
[367]	Test-mae:0.079014
[368]	Test-mae:0.079018
[369]	Test-mae:0.078996
[370]	Test-mae:0.078991
[371]	Test-mae:0.078992
[372]	Test-mae:0.078986
[373]	Test-mae:0.078976
[374]	Test-mae:0.078946
[375]	Test-mae:0.07894
[376]	Test-mae:0.078955
[377]	Test-mae:0.078943
[378]	Test-mae:0.078935
[379]	Test-mae:0.078903
[380]	Test-mae:0.078891
[381]	Test-mae:0.078878
[382]	Test-mae:0.07888
[383]	Test-mae:0.078863
[384]	Test-mae:0.078849
[385]	Test-mae:0.078846
[386]	Test-mae:0.078849
[387]	Test-mae:0.078842
[388]	Test-mae:0.078846
[389]	Test-mae:0.078832
[390]	Test-mae:0.078837
[391]	Test-mae:0.078836
[392]	Test-mae:0.078813
[393]	Test-mae:0.078807
[394]	Test-mae:0.078785
[395]	Test-mae:0.078747
[396]	Test-mae:0.078744
[397]	Test-mae:0.07873
[398]	Test-mae:0.078714
[399]	Test-mae:0.078699
[400]	Test-mae:0.078696
[401]	Test-mae:0.078686
[402]	Test-mae:0.078682
[403]	Test-mae:0.078677
[404]	Test-mae:0.078676
[405]	Test-mae:0.078677
[406]	Test-mae:0.078673
[407]	Test-mae:0.078653
[408]	Test-mae:0.078653
[409]	Test-mae:0.078642
[410]	Test-mae:0.07862
[411]	Test-mae:0.078612
[412]	Test-mae:0.078602
[413]	Test-mae:0.078573
[414]	Test-mae:0.078561
[415]	Test-mae:0.078553
[416]	Test-mae:0.078547
[417]	Test-mae:0.078542
[418]	Test-mae:0.078533
[419]	Test-mae:0.078534
[420]	Test-mae:0.078511
[421]	Test-mae:0.078485
[422]	Test-mae:0.078481
[423]	Test-mae:0.078475
[424]	Test-mae:0.078462
[425]	Test-mae:0.078442
[426]	Test-mae:0.07841
[427]	Test-mae:0.078396
[428]	Test-mae:0.078382
[429]	Test-mae:0.078372
[430]	Test-mae:0.07835
[431]	Test-mae:0.078343
[432]	Test-mae:0.078332
[433]	Test-mae:0.07831
[434]	Test-mae:0.078295
[435]	Test-mae:0.078287
[436]	Test-mae:0.078261
[437]	Test-mae:0.078266
[438]	Test-mae:0.078255
[439]	Test-mae:0.078255
[440]	Test-mae:0.078241
[441]	Test-mae:0.078236
[442]	Test-mae:0.078239
[443]	Test-mae:0.078239
[444]	Test-mae:0.078221
[445]	Test-mae:0.078222
[446]	Test-mae:0.078203
[447]	Test-mae:0.07821
[448]	Test-mae:0.078218
[449]	Test-mae:0.078213
[450]	Test-mae:0.078216
[451]	Test-mae:0.078211
[452]	Test-mae:0.078204
[453]	Test-mae:0.078186
[454]	Test-mae:0.078173
[455]	Test-mae:0.078173
[456]	Test-mae:0.078172
[457]	Test-mae:0.07815
[458]	Test-mae:0.078148
[459]	Test-mae:0.078121
[460]	Test-mae:0.078111
[461]	Test-mae:0.078098
[462]	Test-mae:0.078106
[463]	Test-mae:0.078091
[464]	Test-mae:0.078089
[465]	Test-mae:0.078088
[466]	Test-mae:0.078069
[467]	Test-mae:0.078054
[468]	Test-mae:0.078051
[469]	Test-mae:0.078048
[470]	Test-mae:0.078041
[471]	Test-mae:0.078036
[472]	Test-mae:0.078029
[473]	Test-mae:0.078017
[474]	Test-mae:0.078018
[475]	Test-mae:0.07799
[476]	Test-mae:0.077985
[477]	Test-mae:0.077976
[478]	Test-mae:0.077972
[479]	Test-mae:0.077952
[480]	Test-mae:0.077942
[481]	Test-mae:0.077955
[482]	Test-mae:0.077938
[483]	Test-mae:0.077937
[484]	Test-mae:0.077936
[485]	Test-mae:0.07793
[486]	Test-mae:0.077931
[487]	Test-mae:0.077912
[488]	Test-mae:0.077911
[489]	Test-mae:0.077909
[490]	Test-mae:0.077887
[491]	Test-mae:0.077888
[492]	Test-mae:0.077872
[493]	Test-mae:0.077856
[494]	Test-mae:0.07784
[495]	Test-mae:0.077832
[496]	Test-mae:0.077826
[497]	Test-mae:0.077837
[498]	Test-mae:0.077821
[499]	Test-mae:0.077809
[500]	Test-mae:0.077806
[501]	Test-mae:0.077817
[502]	Test-mae:0.077802
[503]	Test-mae:0.0778
[504]	Test-mae:0.07779
[505]	Test-mae:0.077777
[506]	Test-mae:0.077779
[507]	Test-mae:0.077775
[508]	Test-mae:0.077797
[509]	Test-mae:0.077794
[510]	Test-mae:0.077779
[511]	Test-mae:0.077781
[512]	Test-mae:0.077792
[513]	Test-mae:0.077794
[514]	Test-mae:0.077794
[515]	Test-mae:0.077786
[516]	Test-mae:0.077779
[517]	Test-mae:0.077759
[518]	Test-mae:0.077756
[519]	Test-mae:0.07775
[520]	Test-mae:0.077743
[521]	Test-mae:0.077739
[522]	Test-mae:0.077728
[523]	Test-mae:0.077714
[524]	Test-mae:0.07771
[525]	Test-mae:0.077705
[526]	Test-mae:0.077703
[527]	Test-mae:0.077681
[528]	Test-mae:0.077665
[529]	Test-mae:0.077665
[530]	Test-mae:0.077657
[531]	Test-mae:0.077646
[532]	Test-mae:0.077646
[533]	Test-mae:0.07762
[534]	Test-mae:0.077623
[535]	Test-mae:0.077597
[536]	Test-mae:0.077578
[537]	Test-mae:0.077578
[538]	Test-mae:0.07757
[539]	Test-mae:0.077551
[540]	Test-mae:0.077551
[541]	Test-mae:0.07754
[542]	Test-mae:0.077535
[543]	Test-mae:0.077535
[544]	Test-mae:0.077527
[545]	Test-mae:0.077523
[546]	Test-mae:0.077518
[547]	Test-mae:0.077505
[548]	Test-mae:0.077498
[549]	Test-mae:0.07749
[550]	Test-mae:0.077485
[551]	Test-mae:0.077475
[552]	Test-mae:0.077479
[553]	Test-mae:0.077482
[554]	Test-mae:0.077466
[555]	Test-mae:0.077468
[556]	Test-mae:0.077459
[557]	Test-mae:0.077465
[558]	Test-mae:0.077452
[559]	Test-mae:0.07745
[560]	Test-mae:0.077445
[561]	Test-mae:0.077428
[562]	Test-mae:0.077409
[563]	Test-mae:0.077394
[564]	Test-mae:0.077377
[565]	Test-mae:0.077357
[566]	Test-mae:0.077343
[567]	Test-mae:0.077331
[568]	Test-mae:0.07732
[569]	Test-mae:0.077322
[570]	Test-mae:0.077328
[571]	Test-mae:0.077326
[572]	Test-mae:0.077302
[573]	Test-mae:0.077286
[574]	Test-mae:0.077285
[575]	Test-mae:0.077279
[576]	Test-mae:0.077273
[577]	Test-mae:0.077263
[578]	Test-mae:0.07727
[579]	Test-mae:0.077266
[580]	Test-mae:0.077259
[581]	Test-mae:0.077256
[582]	Test-mae:0.077255
[583]	Test-mae:0.077246
[584]	Test-mae:0.077244
[585]	Test-mae:0.077244
[586]	Test-mae:0.077232
[587]	Test-mae:0.07723
[588]	Test-mae:0.07722
[589]	Test-mae:0.077223
[590]	Test-mae:0.077195
[591]	Test-mae:0.077184
[592]	Test-mae:0.077182
[593]	Test-mae:0.077178
[594]	Test-mae:0.077164
[595]	Test-mae:0.07717
[596]	Test-mae:0.077173
[597]	Test-mae:0.077173
[598]	Test-mae:0.077154
[599]	Test-mae:0.077148
[600]	Test-mae:0.077131
[601]	Test-mae:0.077121
[602]	Test-mae:0.077118
[603]	Test-mae:0.077125
[604]	Test-mae:0.077109
[605]	Test-mae:0.07711
[606]	Test-mae:0.077104
[607]	Test-mae:0.077113
[608]	Test-mae:0.077106
[609]	Test-mae:0.077106
[610]	Test-mae:0.077096
[611]	Test-mae:0.077081
[612]	Test-mae:0.077078
[613]	Test-mae:0.077075
[614]	Test-mae:0.077062
[615]	Test-mae:0.077053
[616]	Test-mae:0.077054
[617]	Test-mae:0.077046
[618]	Test-mae:0.077034
[619]	Test-mae:0.077032
[620]	Test-mae:0.077029
[621]	Test-mae:0.077018
[622]	Test-mae:0.077007
[623]	Test-mae:0.076986
[624]	Test-mae:0.076982
[625]	Test-mae:0.076979
[626]	Test-mae:0.076981
[627]	Test-mae:0.076974
[628]	Test-mae:0.076973
[629]	Test-mae:0.076971
[630]	Test-mae:0.076964
[631]	Test-mae:0.076964
[632]	Test-mae:0.076962
[633]	Test-mae:0.076961
[634]	Test-mae:0.076958
[635]	Test-mae:0.076958
[636]	Test-mae:0.076946
[637]	Test-mae:0.076946
[638]	Test-mae:0.076954
[639]	Test-mae:0.076946
[640]	Test-mae:0.076939
[641]	Test-mae:0.076932
[642]	Test-mae:0.076922
[643]	Test-mae:0.076911
[644]	Test-mae:0.076892
[645]	Test-mae:0.076889
[646]	Test-mae:0.076885
[647]	Test-mae:0.076889
[648]	Test-mae:0.076869
[649]	Test-mae:0.07686
[650]	Test-mae:0.076849
[651]	Test-mae:0.076846
[652]	Test-mae:0.076843
[653]	Test-mae:0.076839
[654]	Test-mae:0.076834
[655]	Test-mae:0.07684
[656]	Test-mae:0.076839
[657]	Test-mae:0.076847
[658]	Test-mae:0.076859
[659]	Test-mae:0.07684
[660]	Test-mae:0.076839
[661]	Test-mae:0.076822
[662]	Test-mae:0.076812
[663]	Test-mae:0.076822
[664]	Test-mae:0.07683
[665]	Test-mae:0.076817
[666]	Test-mae:0.076809
[667]	Test-mae:0.076806
[668]	Test-mae:0.076799
[669]	Test-mae:0.076806
[670]	Test-mae:0.076801
[671]	Test-mae:0.076793
[672]	Test-mae:0.076802
[673]	Test-mae:0.076799
[674]	Test-mae:0.076781
[675]	Test-mae:0.076775
[676]	Test-mae:0.076771
[677]	Test-mae:0.076772
[678]	Test-mae:0.076802
[679]	Test-mae:0.076805
[680]	Test-mae:0.076799
[681]	Test-mae:0.076793
[682]	Test-mae:0.076788
[683]	Test-mae:0.076782
[684]	Test-mae:0.076761
[685]	Test-mae:0.076763
[686]	Test-mae:0.076764
[687]	Test-mae:0.076763
[688]	Test-mae:0.076758
[689]	Test-mae:0.076761
[690]	Test-mae:0.076756
[691]	Test-mae:0.076761
[692]	Test-mae:0.076745
[693]	Test-mae:0.07675
[694]	Test-mae:0.076732
[695]	Test-mae:0.076716
[696]	Test-mae:0.076706
[697]	Test-mae:0.076712
[698]	Test-mae:0.076709
[699]	Test-mae:0.07671
[700]	Test-mae:0.076704
[701]	Test-mae:0.076691
[702]	Test-mae:0.076692
[703]	Test-mae:0.076674
[704]	Test-mae:0.076676
[705]	Test-mae:0.076666
[706]	Test-mae:0.076669
[707]	Test-mae:0.076672
[708]	Test-mae:0.076665
[709]	Test-mae:0.076664
[710]	Test-mae:0.076667
[711]	Test-mae:0.076659
[712]	Test-mae:0.076643
[713]	Test-mae:0.076645
[714]	Test-mae:0.076649
[715]	Test-mae:0.076649
[716]	Test-mae:0.07665
[717]	Test-mae:0.076649
[718]	Test-mae:0.076649
[719]	Test-mae:0.076647
[720]	Test-mae:0.076642
[721]	Test-mae:0.076649
[722]	Test-mae:0.076661
[723]	Test-mae:0.07665
[724]	Test-mae:0.07665
[725]	Test-mae:0.076647
[726]	Test-mae:0.076642
[727]	Test-mae:0.076634
[728]	Test-mae:0.076633
[729]	Test-mae:0.076624
[730]	Test-mae:0.076612
[731]	Test-mae:0.07661
[732]	Test-mae:0.076607
[733]	Test-mae:0.076595
[734]	Test-mae:0.076597
[735]	Test-mae:0.076596
[736]	Test-mae:0.076588
[737]	Test-mae:0.076572
[738]	Test-mae:0.076569
[739]	Test-mae:0.076578
[740]	Test-mae:0.076579
[741]	Test-mae:0.076562
[742]	Test-mae:0.076554
[743]	Test-mae:0.076548
[744]	Test-mae:0.076544
[745]	Test-mae:0.07654
[746]	Test-mae:0.076532
[747]	Test-mae:0.076528
[748]	Test-mae:0.076524
[749]	Test-mae:0.076527
[750]	Test-mae:0.076519
[751]	Test-mae:0.076522
[752]	Test-mae:0.076517
[753]	Test-mae:0.07651
[754]	Test-mae:0.076511
[755]	Test-mae:0.076503
[756]	Test-mae:0.076517
[757]	Test-mae:0.076522
[758]	Test-mae:0.076515
[759]	Test-mae:0.076507
[760]	Test-mae:0.076515
[761]	Test-mae:0.076512
[762]	Test-mae:0.076506
[763]	Test-mae:0.076497
[764]	Test-mae:0.076484
[765]	Test-mae:0.076485
[766]	Test-mae:0.076488
[767]	Test-mae:0.076495
[768]	Test-mae:0.076497
[769]	Test-mae:0.076488
[770]	Test-mae:0.076485
[771]	Test-mae:0.076486
[772]	Test-mae:0.076486
[773]	Test-mae:0.076476
[774]	Test-mae:0.076475
[775]	Test-mae:0.076477
[776]	Test-mae:0.07648
[777]	Test-mae:0.076489
[778]	Test-mae:0.076486
[779]	Test-mae:0.076484
[780]	Test-mae:0.07648
[781]	Test-mae:0.076477
[782]	Test-mae:0.076467
[783]	Test-mae:0.07646
[784]	Test-mae:0.076448
[785]	Test-mae:0.076453
[786]	Test-mae:0.07646
[787]	Test-mae:0.076448
[788]	Test-mae:0.076449
[789]	Test-mae:0.076437
[790]	Test-mae:0.076427
[791]	Test-mae:0.076425
[792]	Test-mae:0.076423
[793]	Test-mae:0.076422
[794]	Test-mae:0.076421
[795]	Test-mae:0.076416
[796]	Test-mae:0.076403
[797]	Test-mae:0.076391
[798]	Test-mae:0.076383
[799]	Test-mae:0.076385
[800]	Test-mae:0.076383
[801]	Test-mae:0.076389
[802]	Test-mae:0.076378
[803]	Test-mae:0.076378
[804]	Test-mae:0.076378
[805]	Test-mae:0.07638
[806]	Test-mae:0.076378
[807]	Test-mae:0.076373
[808]	Test-mae:0.076376
[809]	Test-mae:0.076358
[810]	Test-mae:0.076354
[811]	Test-mae:0.07635
[812]	Test-mae:0.076344
[813]	Test-mae:0.076347
[814]	Test-mae:0.076345
[815]	Test-mae:0.076343
[816]	Test-mae:0.076342
[817]	Test-mae:0.07633
[818]	Test-mae:0.076329
[819]	Test-mae:0.076323
[820]	Test-mae:0.07632
[821]	Test-mae:0.076314
[822]	Test-mae:0.076296
[823]	Test-mae:0.076292
[824]	Test-mae:0.076291
[825]	Test-mae:0.076292
[826]	Test-mae:0.076285
[827]	Test-mae:0.076272
[828]	Test-mae:0.076263
[829]	Test-mae:0.076275
[830]	Test-mae:0.076283
[831]	Test-mae:0.076278
[832]	Test-mae:0.076258
[833]	Test-mae:0.076254
[834]	Test-mae:0.076258
[835]	Test-mae:0.076257
[836]	Test-mae:0.076244
[837]	Test-mae:0.076249
[838]	Test-mae:0.076244
[839]	Test-mae:0.076237
[840]	Test-mae:0.076229
[841]	Test-mae:0.076227
[842]	Test-mae:0.076231
[843]	Test-mae:0.076231
[844]	Test-mae:0.076224
[845]	Test-mae:0.076221
[846]	Test-mae:0.076223
[847]	Test-mae:0.07622
[848]	Test-mae:0.076217
[849]	Test-mae:0.076214
[850]	Test-mae:0.076214
[851]	Test-mae:0.076212
[852]	Test-mae:0.076183
[853]	Test-mae:0.076187
[854]	Test-mae:0.076182
[855]	Test-mae:0.076179
[856]	Test-mae:0.076169
[857]	Test-mae:0.076163
[858]	Test-mae:0.076157
[859]	Test-mae:0.076144
[860]	Test-mae:0.076135
[861]	Test-mae:0.076134
[862]	Test-mae:0.076138
[863]	Test-mae:0.07614
[864]	Test-mae:0.076131
[865]	Test-mae:0.076124
[866]	Test-mae:0.076118
[867]	Test-mae:0.07612
[868]	Test-mae:0.076115
[869]	Test-mae:0.076112
[870]	Test-mae:0.07611
[871]	Test-mae:0.076103
[872]	Test-mae:0.076099
[873]	Test-mae:0.076098
[874]	Test-mae:0.076091
[875]	Test-mae:0.07609
[876]	Test-mae:0.076102
[877]	Test-mae:0.076102
[878]	Test-mae:0.076096
[879]	Test-mae:0.076089
[880]	Test-mae:0.076091
[881]	Test-mae:0.076092
[882]	Test-mae:0.076087
[883]	Test-mae:0.07608
[884]	Test-mae:0.076078
[885]	Test-mae:0.076078
[886]	Test-mae:0.076082
[887]	Test-mae:0.076075
[888]	Test-mae:0.076078
[889]	Test-mae:0.076075
[890]	Test-mae:0.076072
[891]	Test-mae:0.076071
[892]	Test-mae:0.076062
[893]	Test-mae:0.076064
[894]	Test-mae:0.076058
[895]	Test-mae:0.076048
[896]	Test-mae:0.076052
[897]	Test-mae:0.076051
[898]	Test-mae:0.076051
[899]	Test-mae:0.076048
[900]	Test-mae:0.076043
[901]	Test-mae:0.076044
[902]	Test-mae:0.07604
[903]	Test-mae:0.076043
[904]	Test-mae:0.076042
[905]	Test-mae:0.076044
[906]	Test-mae:0.076045
[907]	Test-mae:0.076052
[908]	Test-mae:0.076043
[909]	Test-mae:0.076032
[910]	Test-mae:0.076032
[911]	Test-mae:0.076033
[912]	Test-mae:0.076038
[913]	Test-mae:0.076036
[914]	Test-mae:0.076041
[915]	Test-mae:0.076047
[916]	Test-mae:0.076057
[917]	Test-mae:0.076044
[918]	Test-mae:0.076044
[919]	Test-mae:0.076046
Stopping. Best iteration:
[909]	Test-mae:0.076032

y_pred1 = model.predict(dtest)
y_pred2 = model.predict(dtrain)
mae1 = mean_absolute_error(y_pred1, y_test)
mae2 = mean_absolute_error(y_pred2, y_train)
print('MAE test:', 1 - mae1)
print('MAE train:', 1 - mae2)

MAE test: 0.9239536161775356
MAE train: 0.9579982611245333

dtest_h = xgb.DMatrix(test_X_Headline)

headline = model.predict(dtest_h)

headline

array([-0.06908184,  0.01023591, -0.11381304, ..., -0.09366286,
        0.04553461, -0.04823256], dtype=float32)

Final Model

from sklearn.svm import LinearSVR
from sklearn.metrics import mean_absolute_error

#Title

X_train, X_test, y_train, y_test = train_test_split(train_X_Title, y1, test_size=0.15, random_state=32)
clf1 = LinearSVR(C=0.148,max_iter=20000)
clf1.fit(X_train, y_train)

y_pred1 = clf1.predict(X_test)
mae1 = mean_absolute_error(y_pred1, y_test)
print('MAE for test data :', 1 - mae1)

MAE for test data : 0.9443994994636407

mae1

0.05560050053635933

#Headline
X_train, X_test, y_train, y_test = train_test_split(train_X_Headline, y2, test_size=0.15, random_state=32)
clf2 = LinearSVR(C=0.14,max_iter=20000,tol = 0.00001)
clf2.fit(X_train, y_train)

y_pred2 = clf2.predict(X_test)
y_ = clf2.predict(X_train)
mae2 = mean_absolute_error(y_pred2, y_test)
print('MAE test data:', 1 - mae2)
print('MAE Train',1 - mean_absolute_error(y_, y_train))

MAE test data: 0.9313123759844782
MAE Train 0.954211609292743

/usr/local/lib/python3.6/dist-packages/sklearn/svm/_base.py:947: ConvergenceWarning: Liblinear failed to converge, increase the number of iterations.
  "the number of iterations.", ConvergenceWarning)

mae2

0.06868762401552173

print('MAE:', 1 - ((0.4 * mae1) + (0.6 * mae2)))

MAE: 0.9365472253761432

title = clf1.predict(test_X_Title)
headline = clf2.predict(test_X_Headline)

df = pd.DataFrame()
df['IDLink'] = test_id
df['SentimentTitle'] = title
df['SentimentHeadline'] = headline
df.to_csv('submit_SVR.csv', index=False)