Support Vector Machines

import pandas as pd
import numpy as np
import matplotlib.pyplot as plt
from sklearn.preprocessing import StandardScaler
from sklearn.svm import SVC
from sklearn.model_selection import GridSearchCV

def df_processed(csv_loc):
    """
    Loads and processes the Titanic dataset from a CSV location.

    This function handles data cleaning and feature engineering to prepare
    the data for machine learning. It addresses missing values (NaNs) and
    converts categorical features into numerical format.

    Args:
        csv_loc (str): The file path to the training or test CSV file.

    Returns:
        pd.DataFrame: A processed DataFrame ready for modeling.

    Notes:
        Data Cleaning & Imputation:
        - 'Age': Replaces NaN entries with the median age of the column.
        - 'Fare': Replaces NaN entries with the median fare.
        - 'Embarked': Replaces NaN entries with the mode (most frequent port).

        Feature Engineering & Dropping:
        - 'Cabin', 'Ticket', 'Name': These columns are dropped. 'Cabin' has
          many NaN values, while 'Ticket' and 'Name' are non-numeric object
          types with high cardinality, which are dropped for simplicity.
        - 'Sex' & 'Embarked': These categorical columns are converted into
          numerical dummy variables using one-hot encoding. `drop_first=True`
          is used to create k-1 dummies, which helps avoid
          multicollinearity (e.g., 'Sex' becomes 'Sex_male' with 0/1 values).
    """
    df = pd.read_csv(csv_loc)
    df = df.drop(["Ticket", "Cabin", "Name"], axis=1)
    df["Age"] = df["Age"].fillna(df["Age"].median())
    df["Embarked"] = df["Embarked"].fillna(df["Embarked"].mode())
    df["Fare"] = df["Fare"].fillna(df["Fare"].median()) # 1 passenger in the test set has NaN fare
    df = pd.get_dummies(df, columns=["Embarked", "Sex"], drop_first=True)
    return df

X_tr = df_processed("train.csv").drop(["Survived"], axis=1)
y_tr = df_processed("train.csv")["Survived"]
X_ts = df_processed("test.csv")
scaler = StandardScaler()
X_tr_sc = scaler.fit_transform(X_tr)
X_ts_sc = scaler.transform(X_ts)

svm_model = SVC(C=0.9, kernel='rbf', gamma=1/(X_tr_sc.shape[1]*X_tr_sc.var())) 
svm_model.fit(X_tr_sc, y_tr)
y_ts_hat = svm_model.predict(X_ts_sc)

PassengerIDs = pd.read_csv("test.csv")["PassengerId"].to_numpy()
df_submission = pd.DataFrame({"PassengerId": PassengerIDs, 
                              "Survived": y_ts_hat})
df_submission.to_csv('SVM_submission.csv', index=False)