A Methodology for Validating Diversity in Synthetic Time Series Generation

Abstract

Highlights This paper presents a new method for generating 50K diverse synthetic time series. We present a discussion on time series characteristics and metrics with a view to understanding time series diversity. We developed a robust framework for validating diversity in synthetic time series generation.

MOSY: A method for synthetic opinions to yield a robust fuzzy expert system

“…Related work: Synthetic data generation methods have been considered across distinct domains such as synthetic time series generation [4] , synthetic electronic medical records (of a fairly critical nature) [5] , vehicle image domain randomization [6] , use of Kriging and Radial Basis Functions in Bayesian networks [7] and synthetic keystroke dynamics [8] . In these works, the common motivation was largely the need to compensate for the lack of real-world data.…”

Analyzing Shared Bike Usage Through Graph-Based Spatio-Temporal Modeling

“…It has a network structure where edges model the similarity between two vertices (nodes) in terms of temporal patterns in their activities. This enables the treatment of data as timeseries patterns [17], with the potential to identify trends, seasonal or cyclical components, irregular components and potentially, the diversity [18] within the data. The STGN is defined formally in Def.…”

“…Conversely, a longer interval yields to potentially more similar networks, because aggregates over longer periods may converge to similar values, at the cost of missing out on some of the system evolution behaviours. Yet they are necessary to capture properties such as seasonality and stationarity [18].…”