adityalab / ratss

'Actionable Insights in Multivariate Time-series for Urban Analytics', Anika Tabassum, Supriya Chinthavali, Varisara Tansakul, and B. Aditya Prakash

========================================================================== Paper:

Both paper and supplementary readings are added here. Check Ratss_SigKDD2021_MILETS.pdf

Usage: Note: You need to set the correct MATLAB_path in the makefile (Including the MATLAB executable).

- Example:
    MATLAB_path = '/usr/local/bin/./matlab'

To run Ratss for sample data do as follows,

>> make demo  

'make demo' will run for the sample data (Covid-19 interventions data in the paper) in data/ directory.

Output:

#located in result_file directory -- covid_interventions_exp.csv: nXs contains the rationalization weight of n time-series in s segments in the segment file.

-- covid_interventions_ei.csv: nXs, each column contains the time-series index ranked by rationalization weights in _exp.csv for each segment in the segment file.

-- covid_interventions_Prest.txt: file of n, contains cost Prest of n time-series

-- covid_interventions_ttl_path.txt: contains value of ttl path of the constructed segment graph, i.e., K

-- covid_interventions_B.txt: file of nX3, contains cost K*PB-PRest,PB,Prest of n time-series

Citations:

This paper is under creative common license. If you use our code and paper use the following citations.

@article{tabassum2021actionable, title={Actionable Insights in Multivariate Time-series for Urban Analytics}, author={Tabassum, Anika and Chinthavali, Supriya and Tansakul, Varisara and Prakash, B Aditya}, journal={7th International Workshop of Mining and Learning Time Series in ACM SigKDD}, year={2021} }