TY - GEN
T1 - Distributional impact of customer diversity on hierarchical load forecasting
AU - Zhang, Chi
AU - Li, Ran
AU - Li, Furong
N1 - Publisher Copyright:
© 2020 IEEE.
Copyright:
Copyright 2021 Elsevier B.V., All rights reserved.
PY - 2020/12/16
Y1 - 2020/12/16
N2 - This paper analyzes the distributional impacts of customer diversity on hierarchical load forecasting. Existing research focuses on the group size of the hierarchy but ignores customer diversity within groups such as technologies, tariffs and behaviours. Given the same group size, customer diversity has distributional impacts on forecasting accuracy. For example, higher penetration of photovoltaics could bring more uncertainties while homogeneous customers might reinforce their seasonality, which would widen the spread of errors. This paper utilizes PV and residential datasets from different areas and tariffs to create a pool of diverse profiles. Subsets are sampled and forecasted to study the relationship between customer diversity and probability distributions of forecasting errors. Results indicate that technology brings the greatest impact by altering the mean absolute percentage error maximally by 50.38%, followed by lifestyle (13.60%) and tariffs (5.76%). The overall impacts of customer diversity gradually fade out as the aggregation level (i.e. group size) increases.
AB - This paper analyzes the distributional impacts of customer diversity on hierarchical load forecasting. Existing research focuses on the group size of the hierarchy but ignores customer diversity within groups such as technologies, tariffs and behaviours. Given the same group size, customer diversity has distributional impacts on forecasting accuracy. For example, higher penetration of photovoltaics could bring more uncertainties while homogeneous customers might reinforce their seasonality, which would widen the spread of errors. This paper utilizes PV and residential datasets from different areas and tariffs to create a pool of diverse profiles. Subsets are sampled and forecasted to study the relationship between customer diversity and probability distributions of forecasting errors. Results indicate that technology brings the greatest impact by altering the mean absolute percentage error maximally by 50.38%, followed by lifestyle (13.60%) and tariffs (5.76%). The overall impacts of customer diversity gradually fade out as the aggregation level (i.e. group size) increases.
KW - Customer Diversity
KW - Hierarchical Load Forecasting
KW - Smart Meter Data
UR - http://www.scopus.com/inward/record.url?scp=85099151935&partnerID=8YFLogxK
U2 - 10.1109/PESGM41954.2020.9281634
DO - 10.1109/PESGM41954.2020.9281634
M3 - Conference contribution
AN - SCOPUS:85099151935
T3 - IEEE Power and Energy Society General Meeting
BT - 2020 IEEE Power and Energy Society General Meeting, PESGM 2020
PB - IEEE
CY - U. S. A.
T2 - 2020 IEEE Power and Energy Society General Meeting, PESGM 2020
Y2 - 2 August 2020 through 6 August 2020
ER -