OEDI SI/Scenarios/Data Imputation Scenario

Data Imputation Scenario​ Summary

• Date Created: 2024/01/03
• Organization: ANL
• Objective: Provide estimate for missing streaming measurement data.
• Use Case: Data Preprocessing
• Methodology
• Inputs
• Outputs
• Configuration
• Webinars
Data_preprocessing.pdf

Docker Container

Download the docker container at:

Run Locally

http://localhost:8080/edit_scenario?Data Imputation Scenario

    Run Locally

Input Data

Download input data at:

Output Data

Download output data at:

References

Back to Data Preprocessing

Denoising Autoencoders (DAEs) are unsupervised Deep learning algorithms.

Encoder encodes training into an encoding on a higher or lower dimensional hyperplane.

Decoder decodes the encoding to reconstruct the original data.

The application imputes on streaming data when missingness is detected.

Input from sensor federate

At each time step, if the sensor data has no missing data then the data imputation federate output/publication is the same as sensor federate publication. If there is missing data point(s) then the data imputation federate will publish an estimate for the missing value.

use_oedisi_preprocessor: true

oedisi_preprocessor_federates: "data_imputation"

https://hub.docker.com/r/openenergydatainitiative/oedisi-singlecontainer-demo

System data: https://github.com/openEDI/oedisi-ieee123

User input: https://github.com/openEDI/oedi-si-single-container/blob/main/runner/user_config.json

Depending on the user provided configuration, algorithms will be run and the respective outputs will be generated in csv and feather format.

Instructions on how to access individual run outputs -> https://github.com/openEDI/oedi-si-single-container/tree/main#output