Using explainability tools to inform NILM algorithm performance: a decision tree approach

Rachel Stephen Mollel; Lina Stankovic; Vladimir Stankovic

doi:10.1145/3563357.3566148

Using explainability tools to inform NILM algorithm performance: a decision tree approach

Rachel Stephen Mollel, Lina Stankovic, Vladimir Stankovic

Electronic And Electrical Engineering

Research output: Chapter in Book/Report/Conference proceeding › Conference contribution book

9 Citations (Scopus)

90 Downloads (Pure)

Abstract

Over the years, Non-Intrusive Load Monitoring (NILM) research has focused on improving performance and more recently, generalizing over distinct datasets. However, the trustworthiness of the NILM model itself has hardly been addressed. To this end, it becomes important to provide a reasoning or explanation behind the predicted outcome for NILM models especially as machine learning models for NILM are often treated as black-box models. With this explanation, the models, not only can be improved, but also build trust for wider adoption within various applications. This paper demonstrates how some explainability tools can be used to explain the outcomes of a decision tree multi-classification approach for NILM and how model explainability results in improved feature selection and eventually performance.

Original language	English
Title of host publication	Proceedings of the 9th ACM International Conference on Systems for Energy-Efficient Buildings, Cities, and Transportation
Place of Publication	New York, NY, USA
Publisher	Association for Computing Machinery (ACM)
Pages	368-372
Number of pages	5
ISBN (Print)	9781450398909
DOIs	https://doi.org/10.1145/3563357.3566148
Publication status	Published - 11 Nov 2022
Event	BuildSys'22: Proceedings of the 9th ACM International Conference on Systems for Energy-Efficient Buildings, Cities, and Transportation: 6th International Workshop on Non-Intrusive Load Monitoring - Boston, United States Duration: 9 Dec 2022 → 11 Dec 2022 http://nilmworkshop.org/2022/

Publication series

Name	BuildSys '22
Publisher	Association for Computing Machinery (ACM)

Conference

Conference	BuildSys'22: Proceedings of the 9th ACM International Conference on Systems for Energy-Efficient Buildings, Cities, and Transportation
Country/Territory	United States
City	Boston
Period	9/12/22 → 11/12/22
Internet address	http://nilmworkshop.org/2022/

Keywords

NILM
decision tree
classification
explainability

UN SDGs

This output contributes to the following UN Sustainable Development Goals (SDGs)

Access to Document

10.1145/3563357.3566148

Mollel-etal-NILM-2022-Using-explainability-tools-to-inform-NILM-algorithm-performanceAccepted author manuscript, 2.18 MBLicence: Strath_1

REFIT: Electrical Load Measurements (Cleaned)
Murray, D. (Creator), Stankovic, L. (Supervisor) & Stankovic, V. (Supervisor), University of Strathclyde, 16 Jun 2016
DOI: 10.15129/9ab14b0e-19ac-4279-938f-27f643078cec, http://www.refitsmarthomes.org and 3 more links, http://www.epsrc.ac.uk, http://reshare.ukdataservice.ac.uk/852366/, http://reshare.ukdataservice.ac.uk/852367/ (show fewer)
Dataset

9 Citations
1 Poster
1 Article

Explainability-informed feature selection and performance prediction for nonintrusive load monitoring
Mollel, R. S., Stankovic, L. & Stankovic, V., 17 May 2023, In: Sensors. 23, 10, 26 p., 4845.
Research output: Contribution to journal › Article › peer-review

Open Access
File
7 Citations (Scopus)

38 Downloads (Pure)
Using explainability tools to inform non-intrusive load monitoring algorithm performance: a decision tree approach
Mollel, R. S., Stankovic, L. & Stankovic, V., 24 Nov 2022. 1 p.
Research output: Contribution to conference › Poster

Open Access
File

Cite this

Mollel, R. S., Stankovic, L., & Stankovic, V. (2022). Using explainability tools to inform NILM algorithm performance: a decision tree approach. In Proceedings of the 9th ACM International Conference on Systems for Energy-Efficient Buildings, Cities, and Transportation (pp. 368-372). (BuildSys '22). Association for Computing Machinery (ACM). https://doi.org/10.1145/3563357.3566148

Mollel, Rachel Stephen ; Stankovic, Lina ; Stankovic, Vladimir. / Using explainability tools to inform NILM algorithm performance : a decision tree approach. Proceedings of the 9th ACM International Conference on Systems for Energy-Efficient Buildings, Cities, and Transportation. New York, NY, USA : Association for Computing Machinery (ACM), 2022. pp. 368-372 (BuildSys '22).

@inproceedings{b7893d18f69c4282aaa42ee2b6f1f3ee,

title = "Using explainability tools to inform NILM algorithm performance: a decision tree approach",

abstract = "Over the years, Non-Intrusive Load Monitoring (NILM) research has focused on improving performance and more recently, generalizing over distinct datasets. However, the trustworthiness of the NILM model itself has hardly been addressed. To this end, it becomes important to provide a reasoning or explanation behind the predicted outcome for NILM models especially as machine learning models for NILM are often treated as black-box models. With this explanation, the models, not only can be improved, but also build trust for wider adoption within various applications. This paper demonstrates how some explainability tools can be used to explain the outcomes of a decision tree multi-classification approach for NILM and how model explainability results in improved feature selection and eventually performance.",

keywords = "NILM, decision tree, classification, explainability",

author = "Mollel, \{Rachel Stephen\} and Lina Stankovic and Vladimir Stankovic",

year = "2022",

month = nov,

day = "11",

doi = "10.1145/3563357.3566148",

language = "English",

isbn = "9781450398909",

series = "BuildSys '22",

publisher = "Association for Computing Machinery (ACM)",

pages = "368--372",

booktitle = "Proceedings of the 9th ACM International Conference on Systems for Energy-Efficient Buildings, Cities, and Transportation",

address = "United States",

note = "BuildSys'22: Proceedings of the 9th ACM International Conference on Systems for Energy-Efficient Buildings, Cities, and Transportation : 6th International Workshop on Non-Intrusive Load Monitoring ; Conference date: 09-12-2022 Through 11-12-2022",

url = "http://nilmworkshop.org/2022/",

}

Mollel, RS , Stankovic, L & Stankovic, V 2022, Using explainability tools to inform NILM algorithm performance: a decision tree approach. in Proceedings of the 9th ACM International Conference on Systems for Energy-Efficient Buildings, Cities, and Transportation. BuildSys '22, Association for Computing Machinery (ACM), New York, NY, USA, pp. 368-372, BuildSys'22: Proceedings of the 9th ACM International Conference on Systems for Energy-Efficient Buildings, Cities, and Transportation, Boston, Massachusetts, United States, 9/12/22. https://doi.org/10.1145/3563357.3566148

Using explainability tools to inform NILM algorithm performance: a decision tree approach. / Mollel, Rachel Stephen ; Stankovic, Lina ; Stankovic, Vladimir.
Proceedings of the 9th ACM International Conference on Systems for Energy-Efficient Buildings, Cities, and Transportation. New York, NY, USA: Association for Computing Machinery (ACM), 2022. p. 368-372 (BuildSys '22).

Research output: Chapter in Book/Report/Conference proceeding › Conference contribution book

TY - GEN

T1 - Using explainability tools to inform NILM algorithm performance

T2 - BuildSys'22: Proceedings of the 9th ACM International Conference on Systems for Energy-Efficient Buildings, Cities, and Transportation

AU - Mollel, Rachel Stephen

AU - Stankovic, Lina

AU - Stankovic, Vladimir

PY - 2022/11/11

Y1 - 2022/11/11

N2 - Over the years, Non-Intrusive Load Monitoring (NILM) research has focused on improving performance and more recently, generalizing over distinct datasets. However, the trustworthiness of the NILM model itself has hardly been addressed. To this end, it becomes important to provide a reasoning or explanation behind the predicted outcome for NILM models especially as machine learning models for NILM are often treated as black-box models. With this explanation, the models, not only can be improved, but also build trust for wider adoption within various applications. This paper demonstrates how some explainability tools can be used to explain the outcomes of a decision tree multi-classification approach for NILM and how model explainability results in improved feature selection and eventually performance.

AB - Over the years, Non-Intrusive Load Monitoring (NILM) research has focused on improving performance and more recently, generalizing over distinct datasets. However, the trustworthiness of the NILM model itself has hardly been addressed. To this end, it becomes important to provide a reasoning or explanation behind the predicted outcome for NILM models especially as machine learning models for NILM are often treated as black-box models. With this explanation, the models, not only can be improved, but also build trust for wider adoption within various applications. This paper demonstrates how some explainability tools can be used to explain the outcomes of a decision tree multi-classification approach for NILM and how model explainability results in improved feature selection and eventually performance.

KW - NILM

KW - decision tree

KW - classification

KW - explainability

U2 - 10.1145/3563357.3566148

DO - 10.1145/3563357.3566148

M3 - Conference contribution book

SN - 9781450398909

T3 - BuildSys '22

SP - 368

EP - 372

BT - Proceedings of the 9th ACM International Conference on Systems for Energy-Efficient Buildings, Cities, and Transportation

PB - Association for Computing Machinery (ACM)

CY - New York, NY, USA

Y2 - 9 December 2022 through 11 December 2022

ER -

Mollel RS , Stankovic L , Stankovic V. Using explainability tools to inform NILM algorithm performance: a decision tree approach. In Proceedings of the 9th ACM International Conference on Systems for Energy-Efficient Buildings, Cities, and Transportation. New York, NY, USA: Association for Computing Machinery (ACM). 2022. p. 368-372. (BuildSys '22). Epub 2022 Nov 8. doi: 10.1145/3563357.3566148

Using explainability tools to inform NILM algorithm performance: a decision tree approach

Abstract

Publication series

Conference

Keywords

UN SDGs

Access to Document

Fingerprint

Datasets

REFIT: Electrical Load Measurements (Cleaned)

Research output

Explainability-informed feature selection and performance prediction for nonintrusive load monitoring

Using explainability tools to inform non-intrusive load monitoring algorithm performance: a decision tree approach

Cite this