Let’s say you’re a researcher, data scientist, or analyst and have some time-series data you want to understand what influences it. You can upload it to Likely Spurious and the application will analyze it against our over 1 million datasets, comprised of economic, news ngram, book ngram, health and weather series. We use classic and advanced causal modeling processes, accounting for confounders such as population, inflation, and other patterns which are common causes of spurious associations.
The output is a list of candidate series that may influence your time series or at least provide predictive power. But those aren’t the only use cases. We also provide indicators for cointegration or if you need a proxy series measured at a higher frequency the analysis can usually help with that. While Likely Spurious does not provide you with data we do provide links to the sources.
While providing a descriptive name of the series is not required, if you do, we’ll use generative AI to provide candidates for events that may have caused shocks in your series, such as new regulations, supply chain issues, and market fluctuations.
So when you’re ready to start your time-series research and can use some inspiration, upload your data and within a few hours have a list of potential variables for modeling. But remember, the results are likely spurious.
