id3-classification 0.2.0__tar.gz

id3_classification-0.2.0/PKG-INFO

@@ -0,0 +1,16 @@
+Metadata-Version: 2.4
+Name: id3-classification
+Version: 0.2.0
+Summary: Reusable ID3 Decision Tree Classifier
+Author: nanashi
+License: MIT
+Requires-Python: >=3.8
+Description-Content-Type: text/markdown
+License-File: LICENSE
+Requires-Dist: numpy
+Requires-Dist: pandas
+Dynamic: license-file
+
+# ID3 Classifier
+
+Reusable ID3 Decision Tree Algorithm.

id3_classification-0.2.0/README.md

@@ -0,0 +1,3 @@
+# ID3 Classifier
+
+Reusable ID3 Decision Tree Algorithm.

id3_classification-0.2.0/pyproject.toml

@@ -0,0 +1,28 @@
+[build-system]
+requires = ["setuptools>=68", "wheel"]
+build-backend = "setuptools.build_meta"
+
+[project]
+
+name = "id3-classification"   # MUST be unique worldwide
+version = "0.2.0"
+
+description = "Reusable ID3 Decision Tree Classifier"
+
+readme = "README.md"
+
+requires-python = ">=3.8"
+
+authors = [
+  {name="nanashi"}
+]
+
+license = {text="MIT"}
+
+dependencies = [
+    "numpy",
+    "pandas"
+]
+
+[tool.setuptools.packages.find]
+where = ["src"]

id3_classification-0.2.0/setup.cfg

@@ -0,0 +1,4 @@
+[egg_info]
+tag_build = 
+tag_date = 0
+

id3_classification-0.2.0/src/id3_classification.egg-info/PKG-INFO

@@ -0,0 +1,16 @@
+Metadata-Version: 2.4
+Name: id3-classification
+Version: 0.2.0
+Summary: Reusable ID3 Decision Tree Classifier
+Author: nanashi
+License: MIT
+Requires-Python: >=3.8
+Description-Content-Type: text/markdown
+License-File: LICENSE
+Requires-Dist: numpy
+Requires-Dist: pandas
+Dynamic: license-file
+
+# ID3 Classifier
+
+Reusable ID3 Decision Tree Algorithm.

id3_classification-0.2.0/src/id3_classification.egg-info/SOURCES.txt

@@ -0,0 +1,10 @@
+LICENSE
+README.md
+pyproject.toml
+src/id3_classification.egg-info/PKG-INFO
+src/id3_classification.egg-info/SOURCES.txt
+src/id3_classification.egg-info/dependency_links.txt
+src/id3_classification.egg-info/requires.txt
+src/id3_classification.egg-info/top_level.txt
+src/myid3/__init__.py
+src/myid3/id3.py

id3_classification-0.2.0/src/id3_classification.egg-info/dependency_links.txt

@@ -0,0 +1 @@
+

id3_classification-0.2.0/src/id3_classification.egg-info/requires.txt

@@ -0,0 +1,2 @@
+numpy
+pandas

id3_classification-0.2.0/src/id3_classification.egg-info/top_level.txt

@@ -0,0 +1 @@
+myid3

id3_classification-0.2.0/src/myid3/__init__.py

@@ -0,0 +1 @@
+from .id3 import ID3Classifier

id3_classification-0.2.0/src/myid3/id3.py

@@ -0,0 +1,163 @@
+import numpy as np
+import math
+
+
+class ID3Classifier:
+
+    def __init__(self):
+        self.tree = None
+        self.target = None
+        self.features = None
+
+
+    # ===========================
+    # ENTROPY
+    # ===========================
+    def entropy(self, data):
+
+        values, counts = np.unique(data, return_counts=True)
+
+        ent = 0
+
+        for count in counts:
+
+            prob = count / sum(counts)
+
+            if prob > 0:
+                ent -= prob * math.log2(prob)
+
+        return ent
+
+
+    # ===========================
+    # INFORMATION GAIN
+    # ===========================
+    def information_gain(self, df, feature):
+
+        total_entropy = self.entropy(df[self.target])
+
+        values, counts = np.unique(df[feature], return_counts=True)
+
+        weighted_entropy = 0
+
+        for i in range(len(values)):
+
+            subset = df[df[feature] == values[i]]
+
+            subset_entropy = self.entropy(subset[self.target])
+
+            weight = counts[i] / sum(counts)
+
+            weighted_entropy += weight * subset_entropy
+
+        return total_entropy - weighted_entropy
+
+
+    # ===========================
+    # ID3 TREE BUILDER
+    # ===========================
+    def _id3(self, data, features):
+
+        # Pure node
+        if len(np.unique(data[self.target])) == 1:
+
+            return np.unique(data[self.target])[0]
+
+        # No features left
+        if len(features) == 0:
+
+            return data[self.target].mode()[0]
+
+
+        igs = [self.information_gain(data, f) for f in features]
+
+        best = features[np.argmax(igs)]
+
+        tree = {best: {}}
+
+        remaining = [f for f in features if f != best]
+
+
+        for value in np.unique(data[best]):
+
+            subset = data[data[best] == value]
+
+            if len(subset) == 0:
+
+                tree[best][value] = data[self.target].mode()[0]
+
+            else:
+
+                tree[best][value] = self._id3(subset, remaining)
+
+        return tree
+
+
+    # ===========================
+    # FIT MODEL
+    # ===========================
+    def fit(self, df, features, target):
+
+        self.target = target
+        self.features = features
+
+        self.tree = self._id3(df, features)
+
+
+    # ===========================
+    # PRINT TREE
+    # ===========================
+    def print_tree(self, tree=None, indent=""):
+
+        if tree is None:
+            tree = self.tree
+
+        if not isinstance(tree, dict):
+
+            print(indent + "→", tree)
+
+            return
+
+        for key in tree:
+
+            print(indent + key)
+
+            for value in tree[key]:
+
+                print(indent + " ├─", value)
+
+                self.print_tree(tree[key][value], indent + " │   ")
+
+
+    # ===========================
+    # SINGLE PREDICT
+    # ===========================
+    def _predict(self, tree, sample):
+
+        if not isinstance(tree, dict):
+
+            return tree
+
+        root = list(tree.keys())[0]
+
+        value = sample[root]
+
+        if value not in tree[root]:
+
+            return None
+
+        return self._predict(tree[root][value], sample)
+
+
+    # ===========================
+    # MULTIPLE PREDICT
+    # ===========================
+    def predict(self, df):
+
+        preds = []
+
+        for _, row in df.iterrows():
+
+            preds.append(self._predict(self.tree, row))
+
+        return np.array(preds)