🎥Video 30 min1 / 8

What is Machine Learning?

Machine Learning (ML) is the branch of AI that enables systems to learn from data and improve over time — without explicitly programming every rule.

The Clearest Definition

In traditional programming:

Data + Rules → Results

In machine learning:

Data + Results → Rules (the model)

You feed the model examples, and it derives the rules automatically.

Real Example: Spam Filtering

Traditional approach: Write explicit rules (contains "free" + "click here" → spam). Problem: doesn't cover all cases, needs constant manual updates.

ML approach: Give the model 10,000 emails (5,000 spam + 5,000 real). It learns patterns on its own and discovers spam signals you'd never think of.

Why ML Now?

Three factors aligned: massive data availability, GPU compute power (100x faster than CPU), and breakthrough algorithms (Gradient Descent, Backpropagation, Transformers).

Core Components of Any ML System

Data → Preprocessing → Model → Training → Evaluation → Deployment

Features: Input characteristics fed to the model (house area, rooms, location).

Labels: The correct answer to learn (house price).

Training set: ~80% of data used for learning.

Test set: ~20% of data used to evaluate generalization.

Key Terms

| Term | Definition | |------|----------| | Overfitting | Model memorizes training data but fails on new data | | Underfitting | Model too simple, misses patterns | | Hyperparameter | Settings you tune (learning rate, depth) |

python

import numpy as np
import pandas as pd
from sklearn.datasets import load_iris, load_boston
import matplotlib
matplotlib.use("Agg")  # للتشغيل بدون شاشة
import matplotlib.pyplot as plt

# ─────────────────────────────
# مثال 1: استكشاف بيانات Iris
# ─────────────────────────────
print("=" * 50)
print("مثال 1: بيانات Iris الكلاسيكية")
print("=" * 50)

iris = load_iris()
df = pd.DataFrame(iris.data, columns=iris.feature_names)
df["species"] = [iris.target_names[t] for t in iris.target]

print(f"\nشكل البيانات: {df.shape}")
print(f"الأنواع: {df['species'].unique()}")
print(f"\nأول 5 صفوف:")
print(df.head())

print(f"\nإحصائيات:")
print(df.describe().round(2))

# ─────────────────────────────
# مثال 2: تمييز Features وLabels
# ─────────────────────────────
print("\n" + "=" * 50)
print("مثال 2: فصل Features عن Labels")
print("=" * 50)

X = df[iris.feature_names]   # Features (المدخلات)
y = df["species"]            # Labels (المخرجات)

print(f"Features shape: {X.shape}  ← (عدد الأمثلة, عدد الميزات)")
print(f"Labels shape:   {y.shape}  ← (عدد الأمثلة,)")

# ─────────────────────────────
# مثال 3: Train/Test Split
# ─────────────────────────────
from sklearn.model_selection import train_test_split

X_train, X_test, y_train, y_test = train_test_split(
    X, y, test_size=0.2, random_state=42, stratify=y
)

print(f"\nحجم بيانات التدريب : {X_train.shape[0]} مثال ({X_train.shape[0]/len(X)*100:.0f}%)")
print(f"حجم بيانات الاختبار: {X_test.shape[0]} مثال ({X_test.shape[0]/len(X)*100:.0f}%)")

# ─────────────────────────────
# مثال 4: أول نموذج بسيط
# ─────────────────────────────
from sklearn.neighbors import KNeighborsClassifier
from sklearn.metrics import accuracy_score

print("\n" + "=" * 50)
print("مثال 4: أبسط نموذج — KNeighbors")
print("=" * 50)

model = KNeighborsClassifier(n_neighbors=3)
model.fit(X_train, y_train)

y_pred = model.predict(X_test)
accuracy = accuracy_score(y_test, y_pred)
print(f"\nدقة النموذج على بيانات الاختبار: {accuracy:.1%}")

# تنبؤ بمثال جديد
new_flower = [[5.1, 3.5, 1.4, 0.2]]  # قياسات زهرة جديدة
prediction = model.predict(new_flower)
print(f"\nتنبؤ لزهرة جديدة {new_flower[0]}: {prediction[0]}")