diff --git a/.gitignore b/.gitignore
index 85771a7..46117a8 100644
--- a/.gitignore
+++ b/.gitignore
@@ -2,4 +2,5 @@ venv*
 .venv*
 .vscode*
 __pycache__*
-music_genre.csv
\ No newline at end of file
+music_genre.csv
+music_genre.model
\ No newline at end of file
diff --git a/bayes.py b/bayes.py
index cb7c515..3e155cd 100644
--- a/bayes.py
+++ b/bayes.py
@@ -1,3 +1,35 @@
+from sklearn.naive_bayes import GaussianNB
+from sklearn.metrics import confusion_matrix, accuracy_score
+import pandas as pd
+import numpy as np
+import pickle, os
+import typing
+
 class Bayes:
-    def __init__(self):
-        pass
\ No newline at end of file
+    def __init__(self, path: str):
+        self.path = path
+        self.model_exists = False
+        if os.path.isfile(self.path):
+            self.model_exists = True
+            with open(self.path, 'rb') as file:
+                self.classifier = pickle.load(file)
+        else:
+            self.classifier = GaussianNB()
+
+
+    def train(self, X: pd.DataFrame, Y: pd.Series) -> None:
+        self.classifier.fit(X, Y)
+        with open(self.path, 'wb') as file:
+            pickle.dump(self.classifier, file)
+        self.model_exists = True
+
+
+    def predict(self, X: pd.DataFrame) -> np.ndarray:
+        predictions = self.classifier.predict(X)
+        return predictions
+
+
+    def eval(self, Y: pd.Series, Y_pred: np.ndarray) -> typing.Tuple[np.ndarray, np.float64]:
+        cm = confusion_matrix(Y, Y_pred)
+        ac = accuracy_score(Y, Y_pred)
+        return (cm, ac)
\ No newline at end of file
diff --git a/datapreparator.py b/datapreparator.py
index 9223b22..408a72b 100644
--- a/datapreparator.py
+++ b/datapreparator.py
@@ -1,5 +1,7 @@
+from sklearn.model_selection import train_test_split
 from copy import deepcopy
 import pandas as pd
+import typing
 
 class DataPreparator:
     genre_dict = {
@@ -15,9 +17,16 @@ class DataPreparator:
         "rock" : 10
     }
 
+
     def prepare_data(df: pd.DataFrame) -> pd.DataFrame:
         data = deepcopy(df)
         column = df["label"].apply(lambda x: DataPreparator.genre_dict[x])
         data.insert(0, 'genre', column, 'float')
         data = data.drop(columns=['filename', 'label', 'length'])
-        return data
\ No newline at end of file
+        return data
+
+    
+    def train_test_split(df: pd.DataFrame) -> typing.Tuple[pd.DataFrame, pd.DataFrame, pd.Series, pd.Series]:
+        X = df.drop(["genre"], axis=1)
+        Y = df["genre"]
+        return train_test_split(X, Y, test_size = 0.20, random_state = False)
\ No newline at end of file
diff --git a/main.py b/main.py
index f520e52..be5f612 100644
--- a/main.py
+++ b/main.py
@@ -10,3 +10,20 @@ else:
     data_raw = pd.read_csv('music_genre_raw.csv')
     data = DataPreparator.prepare_data(data_raw)
     data.to_csv(filename, index=False)
+
+X_train, X_test, Y_train, Y_test = DataPreparator.train_test_split(data)
+
+bayes = Bayes('music_genre.model')
+if(not bayes.model_exists):
+    bayes.train(X_train, Y_train)
+
+
+Y_predicted = bayes.predict(X_train)
+eval_result = bayes.eval(Y_train, Y_predicted)
+print("Train:")
+print(eval_result[1])
+
+Y_predicted = bayes.predict(X_test)
+eval_result = bayes.eval(Y_test, Y_predicted)
+print("Test:")
+print(eval_result[1])
diff --git a/requirements.txt b/requirements.txt
index da519a7..20be7de 100644
--- a/requirements.txt
+++ b/requirements.txt
@@ -1 +1,3 @@
-pandas==1.2.4
\ No newline at end of file
+pandas==1.2.4
+numpy==1.20.3
+sklearn==0.0
\ No newline at end of file