From 9728e579d4235c769674b92ae5757482dc71eda8 Mon Sep 17 00:00:00 2001
From: Aleksy Wroblewski <aleksy.wroblewski@bbbit.io>
Date: Sat, 17 Apr 2021 14:23:23 +0200
Subject: [PATCH] Improve the results to 0.8ish

---
 dev-0/out.tsv     |  148 +++---
 solution-dev0.py  |    3 +-
 solution-testA.py |    3 +-
 test-A/out.tsv    | 1282 ++++++++++++++++++++++-----------------------
 4 files changed, 717 insertions(+), 719 deletions(-)

diff --git a/dev-0/out.tsv b/dev-0/out.tsv
index 7881845..311a453 100644
--- a/dev-0/out.tsv
+++ b/dev-0/out.tsv
@@ -1,87 +1,87 @@
-5
-29
-35
-32
-18
-20
-3
-42
-4
-1
-17
-20
-16
-44
-13
-13
-41
-36
-10
+14
 0
-27
-28
-8
-14
-25
-19
-2
-8
-38
-40
-34
-34
-15
-24
-13
-9
-39
-5
-2
-32
 4
-33
-15
-21
-30
-37
-9
-11
-26
-15
-25
-35
-34
-19
-6
-14
-16
-16
-34
 7
-15
+2
+5
+22
 12
-34
-31
-20
-7
-34
-30
-35
+5
+4
+15
+5
+0
+2
+8
+8
+8
+2
 9
-18
-10
-30
-3
-43
-42
-25
+23
+24
 6
+13
+10
+15
+6
+11
+13
+23
+0
+22
+22
+3
+20
+8
+3
+20
+14
+11
+7
+5
+18
+1
+7
+16
 21
 22
-1
+0
+22
+3
+15
+4
+22
+6
 17
+10
+0
+0
+22
+20
+3
+0
+1
+21
+5
+6
+22
+16
+19
+22
+2
+9
+16
 1
 0
+12
 15
+17
+7
+2
+4
+15
+4
+2
+3
+6
 7
-23
diff --git a/solution-dev0.py b/solution-dev0.py
index d611c26..6bad508 100644
--- a/solution-dev0.py
+++ b/solution-dev0.py
@@ -17,12 +17,11 @@ def main():
         documents = [preprocess(document, stopwords)
                      for document in in_file.readlines()]
 
-        vectorizer = TfidfVectorizer(ngram_range=(1, 3), use_idf=False)
         vectorizer = TfidfVectorizer()
 
         document_vectors = vectorizer.fit_transform(documents)
         predictions = KMeans(
-            n_clusters=45, max_iter=1000).fit_predict(document_vectors)
+            n_clusters=25, max_iter=1000).fit_predict(document_vectors)
 
         with open("dev-0/out.tsv", "w") as out_file:
             for prediction in predictions:
diff --git a/solution-testA.py b/solution-testA.py
index 05ac567..3346661 100644
--- a/solution-testA.py
+++ b/solution-testA.py
@@ -17,12 +17,11 @@ def main():
         documents = [preprocess(document, stopwords)
                      for document in in_file.readlines()]
 
-        vectorizer = TfidfVectorizer(ngram_range=(1, 3), use_idf=False)
         vectorizer = TfidfVectorizer()
 
         document_vectors = vectorizer.fit_transform(documents)
         predictions = KMeans(
-            n_clusters=45, max_iter=1000).fit_predict(document_vectors)
+            n_clusters=25, max_iter=1000).fit_predict(document_vectors)
 
         with open("test-A/out.tsv", "w") as out_file:
             for prediction in predictions:
diff --git a/test-A/out.tsv b/test-A/out.tsv
index 15a02f7..fe8b5ad 100644
--- a/test-A/out.tsv
+++ b/test-A/out.tsv
@@ -1,691 +1,691 @@
-39
-17
-11
-27
-41
-2
-39
-12
-37
-31
-38
-34
-12
-42
-23
-30
-14
-36
-14
-23
+18
 3
-43
-2
-19
-22
-21
-38
-5
-12
-16
-25
-12
-2
-11
-32
-4
-34
-13
-11
-9
-37
-44
-27
-12
-22
-42
-40
-7
-11
-23
-11
-44
-16
-19
-44
-13
-16
-11
-16
-6
 1
-27
-10
-41
+5
+3
+11
+4
 6
+18
+3
+14
+17
+6
+20
+22
+3
+9
+5
+9
+22
+13
+0
+11
+20
+20
+18
+5
+4
+2
+18
+10
+6
+11
+1
+19
 24
-19
-2
-27
-18
-25
-32
-14
-44
-26
-39
-40
-6
-37
-6
-22
-8
-33
-16
-41
-34
-13
-1
-38
-37
-0
-25
-32
-41
-27
-27
-39
-22
-38
-42
-14
-43
-14
-44
-31
-40
-4
-5
-36
-38
-12
-11
-6
-11
-26
-41
-41
-38
-25
-40
-3
-12
-44
-40
-39
-11
-14
-13
-12
-38
-25
-32
-1
-13
-5
-32
-20
-4
-33
-20
-44
-42
-3
-20
-21
-42
-38
-20
-1
-13
-14
 17
-27
-23
-19
-10
+4
+1
+6
 14
-12
-31
-43
-19
+4
+9
+6
+23
+23
 15
+21
+1
+22
+1
+4
+15
+20
+13
+10
+3
+1
+18
+18
+15
+9
+15
+9
+13
+1
+20
+11
+10
+3
+10
+19
+9
+4
+18
+4
+10
+18
+20
+13
 13
 7
-25
-41
-33
-0
-32
-32
-38
-4
-26
-44
-14
-33
-2
-42
-13
-15
 8
-3
-3
-35
-21
-21
-12
-28
-12
-4
-18
-21
-10
 13
-12
-32
-40
-11
-25
-39
-37
+3
 17
-39
-40
-14
-27
-11
-7
-34
-39
-12
-34
-23
-38
-7
-34
-12
-2
-23
-6
-23
-13
-25
-35
-28
-32
-0
-36
-33
-10
-16
-5
-13
-7
-2
-25
-32
-19
-34
-40
 18
-6
+14
+3
+18
+20
+10
+19
+3
+9
+16
+18
+18
+18
+12
+12
+0
+9
+4
+2
+18
+13
+2
+13
 24
-5
-38
-41
-4
-4
-42
-7
-40
-2
-8
-12
-17
-28
-9
-42
-0
-3
-13
-12
-14
-1
-20
-25
-40
-38
-17
-12
-42
-23
-25
-3
-44
-31
-42
-12
-3
-31
-43
-25
-26
-3
-12
-9
-3
-11
-14
-2
-36
-40
-36
-38
-34
-4
-27
-9
-27
-5
-7
-36
-44
-4
-17
-42
-1
-17
-14
-38
-31
-4
-31
-9
-33
-16
-25
-44
-18
-21
-17
-33
-43
-16
-17
-37
-38
-26
-25
-42
-16
-33
-1
-13
-17
-11
-38
-31
-23
-38
-38
-1
-43
-23
-19
-3
-14
-34
-13
-27
-19
-7
-25
-12
-14
-36
-4
-14
-44
-30
-23
-32
-10
-23
-4
-38
-11
-9
-7
-0
-38
-37
-10
-5
-12
-12
-13
-31
-19
-36
-21
-22
 6
+1
 22
-39
-12
-22
-10
-29
-2
-21
+1
+17
+3
+3
 24
-1
-11
-23
-31
-39
-41
-14
-4
-16
-32
-13
-12
-19
-23
-29
-12
-1
-12
-37
-9
-12
-40
-13
-25
-7
-4
-1
+18
+17
 21
-33
-36
 6
-2
-0
-17
-11
-41
-34
-23
-32
-19
-7
-44
-8
-1
-44
-12
-19
-2
-16
-11
-37
-13
-14
-15
-30
-8
-12
-42
-13
-44
-7
-12
-29
-13
+4
+10
 3
-7
-13
-4
-25
-40
-44
-15
-9
-14
-11
-40
-44
-37
-37
-39
-37
-39
-38
-42
-17
-6
-27
-23
-0
-28
-17
-38
-22
-35
-40
-4
-17
-38
-44
-11
-7
-40
-12
-16
-12
-20
-41
-3
-16
-11
-38
-25
-2
-30
-43
-14
-12
-22
-2
-19
-2
-38
-17
-9
-15
-14
-38
-7
-12
-44
-0
-36
-16
-44
 1
-33
-12
-12
-8
-36
-37
-11
-21
-43
-44
-23
-4
-40
-11
-20
-41
-29
-12
-2
-22
-12
+9
+10
 6
-12
-30
-37
-17
+18
+10
+19
+15
+10
+18
+19
 5
-32
+14
 8
-39
-18
-32
-24
-2
-11
-26
-12
-3
-30
-13
-44
-12
-32
-11
-11
-25
-15
-37
-34
-12
-1
-34
-16
-13
-3
-36
-2
-39
-36
-8
-13
-3
+5
 4
-15
-26
-38
+17
 21
-43
-38
-18
-1
-1
-44
+5
 11
+5
+15
+5
+15
+10
+9
+9
+3
+22
+20
+17
+9
+6
+10
+0
+12
+5
+15
+21
+10
+15
+8
+20
+19
+19
+24
 23
+16
 4
-23
-13
-26
-2
+9
+8
+10
+15
+10
+24
 7
-29
-42
+21
+18
+17
+19
+18
+6
+8
+6
+13
+18
+11
+17
+3
+18
+19
+5
+1
+18
+18
+0
+16
+18
+23
+9
+9
+1
+21
+3
+18
+2
+17
+18
+15
+14
+17
+6
+11
+2
+20
+22
+15
 10
 12
-39
-44
-34
-21
-3
-7
-16
-38
-35
-43
-4
-11
-1
-28
-35
-35
-14
-35
-34
-7
-32
-16
-34
-39
-33
+8
 19
+12
+13
+8
+17
+5
+18
+10
+13
+11
+0
+19
+20
+16
+5
+18
+22
+1
+14
+18
+18
+13
+13
+5
+21
+18
+24
+7
+6
+16
+8
+6
+20
+12
+21
+10
+6
+9
+15
+5
+2
+5
+24
+16
+18
+5
+18
+10
+18
+4
+4
+10
+6
+21
+9
+4
+10
 23
-43
+24
+6
+5
+5
+1
+9
+11
+18
+18
+20
+24
+22
+13
+9
+5
+9
 23
+17
+18
+4
+13
+13
+15
+15
+5
+3
+18
+3
 18
 3
-13
-12
-0
-36
-19
-5
-4
-4
-42
-8
-34
 6
+8
+24
 13
-27
-17
-34
-44
-26
-2
-36
-38
-16
-12
-12
-44
+18
+18
+18
+20
+8
+0
+3
+5
+15
+3
+23
+18
+20
+18
 14
+15
+10
+10
+1
+24
+3
+14
+18
+2
+15
+6
+22
+20
+21
+18
+17
+10
+22
+18
+11
+10
+2
+9
+18
+13
+9
+18
+9
+23
+4
+17
+3
+13
+18
+1
+6
 21
 12
-14
-39
+3
+4
+17
+17
+6
+2
+10
+9
+20
+3
 19
-39
-33
 13
 4
-25
-2
-38
-32
-2
+18
+18
+6
 0
-12
+17
+20
 11
+18
+1
+15
+1
+18
+2
+18
+3
+9
+2
+18
+18
+10
+6
+20
+22
+18
+6
+15
+6
+18
+5
+6
+18
+10
+13
+21
+13
+5
+15
+18
+18
+18
+11
+20
+16
+1
+3
+16
+22
+14
+20
+21
+18
+7
+15
+4
+6
+20
+11
+18
+1
+20
+2
+9
+24
+3
+7
+2
+17
+16
+4
+18
+6
+18
+10
+17
+21
+15
+13
+16
+17
+4
+24
+6
+9
+1
+18
+4
+14
+15
+14
+18
+18
+9
+14
+16
+10
+19
+2
+12
+8
+4
+24
+4
+22
+15
+13
+16
+3
+18
+1
+21
+20
+6
+18
+6
+5
+3
+21
+18
+1
+24
+18
+14
+3
+0
+18
+6
+4
+11
+20
+18
+24
+16
+14
+24
+9
+17
+21
+6
+3
+12
+18
+3
+4
+17
+8
+6
+6
+7
+18
+14
+1
+11
+0
+23
+22
+13
+14
+1
+5
+9
+18
+2
+11
+3
+6
+13
+6
+3
+20
+16
+18
+19
+7
+18
+18
+19
+1
+11
+1
 23
 6
-39
-44
+18
+18
+10
+18
+6
+19
+1
+1
+13
+24
+16
+17
+6
+18
+17
+18
+10
+18
+18
+11
+16
+18
+7
+10
+18
+13
+24
+18
+20
+4
 0
+2
+18
+15
+15
+4
+1
+22
+13
+22
+10
+23
+18
+18
+18
+5
+17
+6
+18
+4
+17
+4
+18
+21
+4
+3
+10
+0
+13
+1
+15
+8
+23
+13
+9
+18
+17
+21
+14
+3
+18
+18
+8
+20
+2
+18
+22
+18
+5
+10
+6
+20
+13
+20
+20
+24
+10
+4
+7
+17
+18
+10
+9
+16
+17
+20
+23
+11
+14
+24
+3
+6
+6
+4
+11
+15
+6
+9
+18
+12
+14
+8
+3
+4
+18
+11
+18
+19
+11
+12
+6
+1
+22
+0
+18
+4
+12