Time-Series Cross-Validation Splitter | Dataford Interview Questions

Code a custom cross-validation splitter that handles time-series data without leaking future information.

Use strictly increasing timestamps and return chronological train and test index pairs. Each test fold must have a fixed size, may be separated from training by a gap, and may optionally limit training-window size.

Function contract: time_series_split(timestamps, n_splits, test_size, gap=0, max_train_size=None) returns a list of [train_indices, test_indices] pairs.

Constraints

1 <= len(timestamps) <= 10^5
timestamps are strictly increasing
1 <= n_splits
1 <= test_size
0 <= gap
n_splits * test_size + gap <= len(timestamps)
max_train_size is None or max_train_size >= 1

Function Signature

def time_series_split(timestamps, n_splits, test_size, gap=0, max_train_size=None):

Your question is Time-Series Cross-Validation Splitter. Start with the requirements in the Question tab.

Run and submit as often as you like. When you're ready, talk me through your approach or go straight to the code.

Problem

Code a custom cross-validation splitter that handles time-series data without leaking future information.

Function contract: time_series_split(timestamps, n_splits, test_size, gap=0, max_train_size=None) returns a list of [train_indices, test_indices] pairs.

Constraints

1 <= len(timestamps) <= 10^5
timestamps are strictly increasing
1 <= n_splits
1 <= test_size
0 <= gap
n_splits * test_size + gap <= len(timestamps)
max_train_size is None or max_train_size >= 1

Function Signature

def time_series_split(timestamps, n_splits, test_size, gap=0, max_train_size=None):

Your question is Time-Series Cross-Validation Splitter. Start with the requirements in the Question tab.

Run and submit as often as you like. When you're ready, talk me through your approach or go straight to the code.