ctr_prediction.py

"""
Click-Through Rate Prediction with Python: Logistic Regression for E-commerce

Predicts whether a user will click a promotional banner ad based on
behavioral and demographic features.

Workflow:
1. Data exploration
2. Correlation & label distribution
3. Visualization
4. Missing value check
5. Logistic Regression modeling (80:20 train/test split)
6. Evaluation (accuracy, confusion matrix, classification report)
"""

import pandas as pd
import matplotlib.pyplot as plt
import seaborn as sns
from sklearn.model_selection import train_test_split
from sklearn.linear_model import LogisticRegression
from sklearn.metrics import confusion_matrix, classification_report

DATA_URL = "https://storage.googleapis.com/dqlab-dataset/pythonTutorial/ecommerce_banner_promo.csv"

NON_NUMERIC_COLUMNS = ["Ad Topic Line", "City", "Country", "Timestamp"]
TARGET_COLUMN = "Clicked on Ad"


# ---------------------------------------------------------------------------
# 1. DATA EXPLORATION
# ---------------------------------------------------------------------------

def load_data(url: str = DATA_URL) -> pd.DataFrame:
    data = pd.read_csv(url)

    print("\n[1] Data exploration with head(), info(), describe(), shape")
    print("Top 5 rows:")
    print(data.head())
    print("Dataset info:")
    print(data.info())
    print("Descriptive statistics:")
    print(data.describe())
    print("Dataset shape:")
    print(data.shape)

    return data


def explore_correlation_and_distribution(data: pd.DataFrame) -> None:
    print("\n[2] Feature correlation via corr()")
    print(data.corr(numeric_only=True))

    print("\n[3] Label distribution via groupby() and size()")
    print(data.groupby("Clicked on Ad").size())


# ---------------------------------------------------------------------------
# 2. VISUALIZATION
# ---------------------------------------------------------------------------

def plot_age_histogram(data: pd.DataFrame) -> None:
    sns.set_style("whitegrid")
    plt.style.use("fivethirtyeight")

    plt.figure(figsize=(10, 5))
    plt.hist(data["Age"], bins=data.Age.nunique())
    plt.xlabel("Age")
    plt.tight_layout()
    plt.show()


def plot_feature_pairplot(data: pd.DataFrame) -> None:
    plt.figure()
    sns.pairplot(data)
    plt.show()


# ---------------------------------------------------------------------------
# 3. MISSING VALUE CHECK
# ---------------------------------------------------------------------------

def check_missing_values(data: pd.DataFrame) -> None:
    print("\n[5] Missing value check")
    print(data.isnull().sum().sum())


# ---------------------------------------------------------------------------
# 4. MODELING
# ---------------------------------------------------------------------------

def build_features_and_target(data: pd.DataFrame):
    """Drop non-numeric columns (Logistic Regression requires numeric input)
    and split into features (X) and target (y)."""
    X = data.drop(NON_NUMERIC_COLUMNS + [TARGET_COLUMN], axis=1)
    y = data[TARGET_COLUMN]
    return X, y


def train_model(X, y, test_size: float = 0.2, random_state: int = 42):
    """Split data 80:20 and train a Logistic Regression classifier."""
    X_train, X_test, y_train, y_test = train_test_split(
        X, y, test_size=test_size, random_state=random_state
    )

    logreg = LogisticRegression()
    logreg = logreg.fit(X_train, y_train)
    y_pred = logreg.predict(X_test)

    print("\n[6] Logistic Regression, 80:20 train/test split")
    print("Model evaluation:")
    print("Training Accuracy:", logreg.score(X_train, y_train))
    print("Testing Accuracy :", logreg.score(X_test, y_test))

    return logreg, X_train, X_test, y_train, y_test, y_pred


# ---------------------------------------------------------------------------
# 5. EVALUATION
# ---------------------------------------------------------------------------

def print_confusion_matrix_and_report(y_test, y_pred) -> None:
    print("\n[7] Confusion matrix and classification report")

    print("Confusion matrix:")
    cm = confusion_matrix(y_test, y_pred)
    print(cm)

    print("Classification report:")
    cr = classification_report(y_test, y_pred)
    print(cr)


def plot_confusion_matrix(y_test, y_pred) -> None:
    cm = confusion_matrix(y_test, y_pred)

    plt.figure(figsize=(6, 5))
    sns.heatmap(
        cm, annot=True, fmt="d",
        cmap=sns.color_palette(["#800000", "#004080"]),
        cbar=False, linewidths=1, linecolor="white"
    )

    plt.title("Confusion Matrix", fontsize=16, color="#800000", weight="bold")
    plt.xlabel("Predicted Label", fontsize=12, color="#004080")
    plt.ylabel("True Label", fontsize=12, color="#004080")

    plt.xticks(ticks=[0.5, 1.5], labels=["0", "1"], fontsize=11)
    plt.yticks(ticks=[0.5, 1.5], labels=["0", "1"], fontsize=11, rotation=0)

    plt.tight_layout()
    plt.show()


def plot_classification_report_table(y_test, y_pred) -> None:
    report = classification_report(y_test, y_pred, output_dict=True)
    df = pd.DataFrame(report).transpose()

    # Total support for numeric classes only (0, 1)
    total_support = df.loc[["0", "1"], "support"].sum()
    df.loc["accuracy", "support"] = total_support

    fig, ax = plt.subplots(figsize=(8, 3))
    ax.axis("off")
    ax.axis("tight")

    table = ax.table(
        cellText=df.round(2).values,
        colLabels=df.columns,
        rowLabels=df.index,
        loc="center",
        cellLoc="center"
    )

    for (i, j), cell in table.get_celld().items():
        cell.set_fontsize(10)
        if i == 0 or j == -1:
            cell.set_text_props(weight="bold", color="white")
            cell.set_facecolor("#800000")
        elif i % 2 == 0:
            cell.set_facecolor("#f7e6e6")
        else:
            cell.set_facecolor("#f2dcdc")

    table.scale(1.2, 1.2)
    plt.title("Classification Report", fontsize=14, color="#800000", weight="bold")
    plt.show()


# ---------------------------------------------------------------------------
# MAIN
# ---------------------------------------------------------------------------

def main():
    data = load_data()
    explore_correlation_and_distribution(data)

    plot_age_histogram(data)
    plot_feature_pairplot(data)

    check_missing_values(data)

    X, y = build_features_and_target(data)
    logreg, X_train, X_test, y_train, y_test, y_pred = train_model(X, y)

    print_confusion_matrix_and_report(y_test, y_pred)
    plot_confusion_matrix(y_test, y_pred)
    plot_classification_report_table(y_test, y_pred)


if __name__ == "__main__":
    main()