Back to Problems

Cross-Validation Data Split Implementation (medium)

Write a Python function that performs k-fold cross-validation data splitting from scratch. The function should take a dataset (as a 2D NumPy array where each row represents a data sample and each column represents a feature) and an integer k representing the number of folds. The function should split the dataset into k parts, systematically use one part as the test set and the remaining as the training set, and return a list where each element is a tuple containing the training set and test set for each fold.

Example

Example:
        input: data = np.array([[1, 2], [3, 4], [5, 6], [7, 8], [9, 10]]), k = 5
        output: [[[[3, 4], [5, 6], [7, 8], [9, 10]], [[1, 2]]],
                [[[1, 2], [5, 6], [7, 8], [9, 10]], [[3, 4]]],
                [[[1, 2], [3, 4], [7, 8], [9, 10]], [[5, 6]]], 
                [[[1, 2], [3, 4], [5, 6], [9, 10]], [[7, 8]]], 
                [[[1, 2], [3, 4], [5, 6], [7, 8]], [[9, 10]]]]
        reasoning: The dataset is divided into 5 parts, each being used once as a test set while the remaining parts serve as the training set.

Understanding k-Fold Cross-Validation Data Splitting

k-Fold cross-validation is a technique used to evaluate the generalizability of a model by dividing the data into `k` folds or subsets. Each fold acts as a test set once, with the remaining `k-1` folds serving as the training set. This approach ensures that every data point gets used for both training and testing, improving model validation.

Steps in k-Fold Cross-Validation Data Split:

  1. Shuffle the dataset randomly. (but not in this case because we test for a unique result)
  2. Split the dataset into k groups.
  3. Generate Data Splits: For each group, treat that group as the test set and the remaining groups as the training set.

Benefits of this Approach:

- Ensures all data is used for both training and testing. - Reduces bias since each data point gets to be in a test set exactly once. - Provides a more robust estimate of model performance. Implementing this data split function will allow a deeper understanding of how data partitioning affects machine learning models and will provide a foundation for more complex validation techniques.
import numpy as np

def cross_validation_split(data: np.ndarray, k: int, seed=42) -> list:
    np.random.seed(seed)
    np.random.shuffle(data)

    n, m = data.shape
    sub_size = int(np.ceil(n / k))
    id_s = np.arange(0, n, sub_size)
    id_e = id_s + sub_size
    if id_e[-1] > n: id_e[-1] = n

    return [[np.concatenate([data[: id_s[i]], data[id_e[i]:]], axis=0).tolist(), data[id_s[i]: id_e[i]].tolist()] for i in range(k)]

There’s no video solution available yet 😔, but you can be the first to submit one at: GitHub link.

Your Solution

Output will be shown here.