From 9dfe8fe3b582808a6df40317dfd5d972afb85264 Mon Sep 17 00:00:00 2001
From: =?UTF-8?q?=C5=81ukasz=20J=C4=99dyk?= <lukasz.jedyk@gmail.com>
Date: Wed, 12 May 2021 21:22:41 +0200
Subject: [PATCH] ver 2 10000 train

---
 dev-0/out.tsv  | 20 +++++++++++---------
 main.py        |  8 ++++++--
 test-A/out.tsv | 23 ++++++++++++-----------
 3 files changed, 29 insertions(+), 22 deletions(-)

diff --git a/dev-0/out.tsv b/dev-0/out.tsv
index 92eb1fa..d328e94 100644
--- a/dev-0/out.tsv
+++ b/dev-0/out.tsv
@@ -2399,7 +2399,7 @@
 1882.9509376976243
 2006.2831425999086
 1901.8236927643368
-1845.8431605627075
+1846.2982848835586
 1907.4807609148313
 1919.1770226218323
 1808.6714849322975
@@ -4770,7 +4770,7 @@
 1869.227930644483
 1906.3844663610223
 1945.3424395378358
-1897.9459278201575
+1897.8191028561196
 1902.4248382081591
 1916.941205362806
 1951.3650770196946
@@ -6336,7 +6336,8 @@
 1974.3692055493452
 1851.1544441272165
 1837.2440449549101
-1921.004623698161
+1906.469703262039
+1941.217890583569
 1863.050896361201
 1935.4381446070022
 1838.6591222095797
@@ -11780,7 +11781,7 @@
 1971.261305356721
 1965.7895878732024
 1894.1926433945762
-1843.5769117112493
+1843.4054815753093
 1798.5510049748643
 1874.6762363908592
 1933.9946144817523
@@ -11868,7 +11869,7 @@
 1933.1659502110472
 1948.6097367533557
 1892.0990307008867
-1881.7539718505154
+1882.4301864075958
 1898.8163564591582
 1891.7720648916354
 1899.1896641539054
@@ -13607,7 +13608,8 @@
 1899.4677481479275
 1832.0157188177106
 1830.994476738317
-1859.8255318357274
+1872.857763141983
+1887.1141759579975
 1821.8836244133079
 1913.4663037228643
 1985.0371631137966
@@ -14150,7 +14152,7 @@
 1958.9030832469919
 1917.5848215968706
 1905.0147395763986
-1871.6714142048124
+1871.3321978094336
 1901.0391754660595
 1871.966900926708
 1965.8106578756367
@@ -17368,7 +17370,7 @@
 1863.7139407495256
 1980.5486905106713
 1820.0882670928129
-1856.3160414564259
+1855.6257391640597
 1921.6397748707468
 1970.804687384616
 1902.8181073379312
@@ -19795,7 +19797,7 @@
 1917.3208523541055
 1850.9065744029447
 1912.2333094584328
-1837.7607134713
+1837.590902324513
 1969.993000236959
 1956.0008619219907
 1941.4140218594555
diff --git a/main.py b/main.py
index b0f27c1..b0e1dad 100644
--- a/main.py
+++ b/main.py
@@ -1,3 +1,4 @@
+import csv
 import pandas as pd
 from sklearn.linear_model import LinearRegression
 from sklearn.pipeline import make_pipeline
@@ -6,8 +7,8 @@ from sklearn.feature_extraction.text import TfidfVectorizer
 col_names = ['start_date', 'end_date', 'title', 'source', 'content']
 
 train_set = pd.read_table('train/train.tsv.xz', error_bad_lines=False, header=None, names=col_names)
-dev_set = pd.read_table('dev-0/in.tsv', error_bad_lines=False, header=None, names=col_names[4:])
-test_set = pd.read_table('test-A/in.tsv', error_bad_lines=False, header=None, names=col_names[4:])
+dev_set = pd.read_table('dev-0/in.tsv', error_bad_lines=False, header=None, names=col_names[4:], quoting=csv.QUOTE_NONE,)
+test_set = pd.read_table('test-A/in.tsv', error_bad_lines=False, header=None, names=col_names[4:], quoting=csv.QUOTE_NONE,)
 
 train_set = train_set.head(10000)
 
@@ -17,6 +18,9 @@ y_train = (train_set['start_date'] + train_set['end_date']) / 2
 X_dev = dev_set['content']
 X_test = test_set['content']
 
+print(dev_set)
+print(test_set)
+
 print('Trenowanie modelu...')
 model = make_pipeline(TfidfVectorizer(), LinearRegression())
 model.fit(X_train, y_train)
diff --git a/test-A/out.tsv b/test-A/out.tsv
index dade4f0..1c618ad 100644
--- a/test-A/out.tsv
+++ b/test-A/out.tsv
@@ -206,7 +206,7 @@
 1928.6587695672113
 1990.9156424576067
 1881.4836785253315
-1884.4519979671402
+1884.046089520049
 1877.4040287725156
 1911.8947608148899
 1899.7628344836432
@@ -961,7 +961,7 @@
 1920.806099260425
 1934.7567428867874
 2000.6897694955856
-1914.538168494774
+1915.0424831770722
 1961.5348637593509
 1976.0306055365545
 1990.438972379044
@@ -1458,7 +1458,7 @@
 1897.0381253809762
 1898.2169614742145
 1907.1817765399285
-1899.7781849186872
+1900.5051189207925
 1916.2254375826196
 1887.7589605584237
 1900.9036082026578
@@ -1944,7 +1944,7 @@
 1932.4369032498905
 1924.7608021483702
 1932.6665753305242
-1910.2637014903585
+1911.1136741658142
 1844.3306216770322
 1920.9735284228261
 1898.4191575441814
@@ -3448,7 +3448,8 @@
 1985.828576031535
 1929.0980372984723
 1963.416770426545
-1983.403262556507
+1958.0061961875651
+1979.5713424319551
 1927.054394299714
 1887.5844594202247
 1952.952473384423
@@ -5241,7 +5242,7 @@
 1961.7395028010487
 1911.578177275467
 1993.361626034721
-1860.938964267848
+1861.582235755756
 1880.8873010116863
 1891.0584427244678
 1927.9286642737504
@@ -6326,7 +6327,7 @@
 1894.4410835153267
 1961.8656695163497
 1918.6540411919113
-1902.5162709462134
+1901.772008274914
 1970.120121368801
 1958.8069602196501
 1822.1082097880283
@@ -8450,7 +8451,7 @@
 1937.635613281464
 1931.6145334405564
 1911.15006803172
-1937.8085829935567
+1937.807751149183
 1911.5828102456112
 1923.676307918832
 1934.4279819781618
@@ -10178,7 +10179,7 @@
 1923.6971313843694
 1914.7503506572486
 1958.747533690674
-1917.7393715934418
+1918.0059649038217
 1909.3267326347955
 1982.6082963993645
 1963.8563866041861
@@ -11287,7 +11288,7 @@
 1930.5405282461093
 1855.5220302615335
 1952.8298961854482
-1893.2863705412592
+1892.6402597321664
 1928.501989244598
 1967.1929378220514
 1911.8069275316302
@@ -13417,7 +13418,7 @@
 2002.918285090558
 1907.5419986831682
 1982.5755707544743
-1844.5958529536363
+1844.823717363196
 1940.7293100159388
 1939.6370593406318
 1910.3139645164986