The SUMSARIZER tool helps researchers gain insight into usage of cooking stoves from data collected by SUMS by faster and more accurate transformation of raw temperature readings into cooking events.
“The Stove Use Monitoring System (SUMS) provides information on adoption of new technologies and their effect on cooking habits. SUMS are based on low-cost commercially available temperature loggers coupled with our fit-for-purpose processing software developed with the Kirk Smith Research Group at the University of California-Berkeley. By recording stove temperature, SUMS provides insights into usage patterns, number of meals cooked and time of use.”
The raw data that SUMs collect is not that useful by itself: it must be analyzed to determine when someone is cooking and for how long. This turns out to be difficult to automate and tedious to do by hand.
The SUMSARIZER uses machine learning to turn a small amount of work into a full analysis of a data set.
Researchers upload SUMs data files for a stove study, then the application prompts the labelling of a small subset of the data. The machine learning algorithms learn from this hand-labelled data and label the rest automatically.
The result is an easy-to-use data set of cooking events, ready for insightful analysis.
SUMSARIZER was the brainchild of three Berkeley researchers
The web application was architected and is currently maintained by
The full source for the SUMSARIZER application is available on Github. We welcome pull requests and forks. Let us know how we can help you make use of SUMs.
Please feel free to submit feature requests using Github Issues.
If you have questions, contact the current maintainer: email@example.com