Implement a k-nearest neighbors algorithm.

To implement k-Nearest Neighbors (k-NN) efficiently, we can leverage NumPy for vectorized operations. First, calculate the Euclidean distances between the new data point (X_new) and all training points (Xtrain) using np.linalg.norm. This avoids explicit loops. Then, instead of sorting all distances, use np.argpartition(distances, k)[:k] to find the indices of the k smallest distances in linear time complexity. Finally, to predict the class for X_new, determine the majority class among the k nearest neighbors by summing their corresponding labels (ytrain) and checking if the sum exceeds k / 2.0.