timeseriesflattener review
A focused Python library for turning irregular time series into prediction-ready tables.
Reviewed by iTechGuides Editors · Editorial team · Updated Oct 2026
Irregularly sampled records can be difficult to shape for machine-learning prediction. timeseriesflattener is an open-source Python package that converts time-series data, including electronic health records, into tabular dataframes. It is aimed at researchers and small teams constructing features or outcomes around specified prediction times, rather than people seeking a general-purpose data-preparation suite. Its central workflow groups observations by an entity ID and produces one row for each prediction time.
Feature construction is the package’s clearest strength. Users can specify lookbehind or lookahead windows, choose aggregation functions such as mean, maximum and minimum, and set fallback values for windows with no observations. Multiple feature specifications can be combined, including temporal predictors, temporal outcomes and static predictors. The package can also generate predictors from embedded text features. This makes it suited to workflows where the timing of observations relative to a prediction matters; the focus is narrower than broad data cleaning or preparation across many kinds of datasets.
The package is installed locally with pip and uses self-hosted deployment, so it fits Python-based workflows that can manage their own environment. Documentation includes tutorials and API reference pages, while support channels are community and documentation. It is open source and has a free plan; there are no commercial tiers described here to compare. Choose timeseriesflattener when you need to convert irregular, entity-linked time series into prediction-time tables with configurable temporal windows. If your priority is hosted deployment, a wider data-preparation toolkit or vendor support beyond community and docs, this focused package is a less suitable fit.
timeseriesflattener pros and cons
- Where it wins
- Creates prediction-time rows and aggregates values by entity ID
- Configurable lookbehind and lookahead windows with fallback values
- Supports temporal and static specifications, including embedded text features
- Where it doesn't
- Requires local installation with pip and self-hosted deployment
- Focused on time-series modeling rather than broad data preparation
- Support is through documentation and the community
timeseriesflattener fact sheet, pricing and score →
Advertiser disclosure: iTechGuides is reader-supported. Vendors can pay for top positions in our rankings and for a place on other products' pages, and we may earn a commission when you click some links. Bought places stand at the number bought; this page and our disclosure say so. Paying vendors may get a fuller, more detailed review from our editors, and that review can change their score. How we rank.
Last updated · How we research and update