Missingness evaluation and analysis strategies for brief and lengthy sequences imputation

Time is probably the most well-defined continuum in physics and, therefore, in nature. It must be of no shock, then, the significance of continuity in time collection datasets — a chronological sequence of observations.

This idea alone drives the motivation behind this text. Actual-world datasets are inclined to lacking values for numerous causes, akin to defective sensors, failures in knowledge ingestion, or just the absence of knowledge throughout a given time. That, nevertheless, doesn’t change the underlying nature of the data-generating strategy of your options.

Understanding what brought about these interruptions and analyzing and dealing with them in a time collection dataset is, due to this fact, paramount to any subsequent job.

Desk of Contents

The Objective of this Article

After a complete exploratory evaluation of your time collection, you may discover that lacking values are current to a substantial extent. By looking for an understanding of the character of your knowledge, it’s best to be capable to differentiate a niche that represents missingness from a niche that entails an precise interruption, characterizing it as an intermittent collection.

This text will give attention to the primary state of affairs — evaluation of lacking values and strategies to judge imputation outcomes. Whereas the precise strategies to carry out imputation are many [1][2], I’ll elaborate on the…