Real-world time series often contain missing values due to sensor failures, power outages or transmission problems. The recovery of those missing values allows better analysis of time series. Several methods have been proposed to recover missing values in time series, which can be algebraic, statistical, machine learning, etc. Selecting the best recovery method highly depends on the time series features, the missing rate and type, and error metric. ImputeBench is a popular terminal-based benchmark that compares over 15 missing value imputation techniques algorithms. It relies on a reusable code framework, a large selection of representative time series, and mechanisms to automate benchmarking. Prediction in time series is a longstanding problem. Several prediction techniques have been proposed in the literature, each showing superior results in some use cases. However, little is known about their relative performance, as existing comparisons are limited to either a small subset of relevant algorithms or to very few datasets or often both. Drawing general conclusion about the performance of prediction techniques remains a challenge.

The goal of the thesis is to construct a helper graphical tool aimed at visualizing different aspects of missing value recovery process - parametrization, patterns of missing values, time series features, etc. more…