smoothing

run.py

@@ -3,50 +3,53 @@ import csv
 import regex as re
 from nltk import bigrams, word_tokenize
 from collections import Counter, defaultdict
-import string
-import unicodedata
-
-data = pd.read_csv(
-    "train/in.tsv.xz",
-    sep="\t",
-    error_bad_lines=False,
-    header=None,
-    quoting=csv.QUOTE_NONE,
-)
-train_labels = pd.read_csv(
-    "train/expected.tsv",
-    sep="\t",
-    error_bad_lines=False,
-    header=None,
-    quoting=csv.QUOTE_NONE,
-)
-
-train_data = data[[6, 7]]
-train_data = pd.concat([train_data, train_labels], axis=1)
-
-train_data["final"] = train_data[6] + train_data[0] + train_data[7]
-
-model = defaultdict(lambda: defaultdict(lambda: 0))
 
 
 def clean(text):
     text = str(text).lower().replace("-\\n", "").replace("\\n", " ")
     return re.sub(r"\p{P}", "", text)
 
-def train_model(data):
+
+class Collection:
+    def __init__(self, path: str) -> None:
+        self._path = path
+
+    def read(self, nrows=200_000):
+        self.data = pd.read_csv(
+            self._path,
+            sep="\t",
+            error_bad_lines=False,
+            header=None,
+            quoting=csv.QUOTE_NONE,
+            nrows=nrows,
+        )
+
+
+class Model:
+    def __init__(self, alpha: float = 0.01) -> None:
+        self.alpha = alpha
+        self.model = defaultdict(lambda: defaultdict(lambda: 0))
+        self.vocab = set()
+
+    def train(self, data):
         for _, row in data.iterrows():
             words = word_tokenize(clean(row["final"]))
         for w1, w2 in bigrams(words, pad_left=True, pad_right=True):
             if w1 and w2:
-                model[w1][w2] += 1
-    for w1 in model:
-        total_count = float(sum(model[w1].values()))
-        for w2 in model[w1]:
-            model[w1][w2] /= total_count
+                self.model[w1][w2] += 1
+                self.vocab.add(w1)
+                self.vocab.add(w2)
 
+        for w1 in self.model:
+            total_count = float(sum(self.model[w1].values()))
+            for w2 in self.model[w1]:
+                self.model[w1][w2] /= total_count
+                self.model[w1][w2] = (self.model[w1][w2] + self.alpha) / (
+                    total_count + self.alpha * len(self.vocab)
+                )
 
-def predict(word):
-    predictions = dict(model[word])
+    def _predict(self, word):
+        predictions = dict(self.model[word])
         most_common = dict(Counter(predictions).most_common(5))
 
         total_prob = 0.0
@@ -66,21 +69,40 @@ def predict(word):
 
         return str_prediction
 
-
-def predict_data(read_path, save_path):
-    data = pd.read_csv(
-        read_path, sep="\t", error_bad_lines=False, header=None, quoting=csv.QUOTE_NONE
-    )
+    def _save(self, save_path: str, data):
         with open(save_path, "w") as file:
             for _, row in data.iterrows():
                 words = word_tokenize(clean(row[6]))
                 if len(words) < 3:
                     prediction = "the:0.2 be:0.2 to:0.2 of:0.1 and:0.1 a:0.1 :0.1"
                 else:
-                prediction = predict(words[-1])
+                    prediction = self._predict(words[-1])
                 file.write(prediction + "\n")
 
+    def predict(self, read_path: str, save_path: str):
+        collection = Collection(read_path)
+        collection.read()
+        self._save(save_path, collection.data)
 
-train_model(train_data)
-predict_data("dev-0/in.tsv.xz", "dev-0/out.tsv")
-predict_data("test-A/in.tsv.xz", "test-A/out.tsv")
+
+
+if __name__ == '__main__':
+
+    data = Collection("train/in.tsv.xz")
+    data.read()
+
+    train_labels = Collection("train/expected.tsv")
+    train_labels.read()
+
+
+    train_data = data.data[[6, 7]]
+    train_data = pd.concat([train_data, train_labels.data], axis=1)
+
+    train_data["final"] = train_data[6] + train_data[0] + train_data[7]
+
+
+    model = Model()
+    model.train(train_data)
+
+    model.predict("dev-0/in.tsv.xz", "dev-0/out.tsv")
+    model.predict("test-A/in.tsv.xz", "test-A/out.tsv")