We can use a custom iterative sampling such that we continuously add samples to the train set. We only should keep in mind that the sample used for validation should be added to the next train sets and a new sample is used for validation.
When working with time-series data in machine learning, the most suitable sampling technique is often “time-based splitting” or “time-series splitting.” Time-series data has a temporal structure, where the order of observations matters. Therefore, randomly shuffling the data or using traditional random sampling techniques may lead to data leakage and incorrect model evaluation.
Time-series splitting involves dividing the dataset into training and testing sets based on the chronological order of the observations. Typically, earlier data points are used for training, and later data points are used for testing. This approach helps the model learn from past data and evaluate its performance on future data, simulating the real-world scenario where the model needs to make predictions on unseen future observations.
In Python, the TimeSeriesSplit class from the scikit-learn library can be used to implement time-series splitting for cross-validation.