Fahad Mehmood
Current Organizations / Institutes
The organizations/institutes from the scholar's latest article
Previous Organizations / Institutes
The organizations / institutes from the scholar's other articles in the past ten years
Fahad Mehmood has a total of 45 co-authors. Here are the top co-authors by number of shared publications:
Hussain Kazmi
5 shared publications- Hussain Kazmi
KU Leuven
4 shared publications - Tahir Abbas Syed
Alliance Manchester Business School, University of Manchester, Manchester, UK
3 shared publications - Johan Driesen
ESAT/ELECTA KU Leuven, Heverlee, Belgium
2 shared publications - Yasir Tariq Mohmand
Department of Management Sciences, COMSATS University Islamabad, Islamabad, Pakistan
2 shared publications Zainab Riaz
2 shared publications
Latest Publications
- Open Access
- 22 August 2026
- Published by Elsevier in SSRN Electronic Journal
Not Adding Up: Bridging Coherency Gaps in Energy Forecasts Using Hierarchical Clustering and Reconciliation
Accurate forecasts of distributed energy resource output are needed at multiple aggregation levels for grid operation and flexibility scheduling. Energy service companies forecast at the unit level while grid operators forecast at the feeder level, independently. Summing unit forecasts does not reproduce the feeder forecast, after accounting for losses. This inconsistency, the coherency gap, distorts estimates of demand or output and the flexibility needs that follow. We quantify this gap on two real-world datasets: residential-demand from 2,072 UK households and industrial combined heat and power generation from 82 Belgian units. Forecasts disagree by 3.5% to 8.8% of total load on average, growing with unit-level heterogeneity. Hierarchical reconciliation enforces aggregation consistency, but in shallow two-level hierarchies it yields only marginal accuracy gains, aggregation structure, not reconciliation methodology, governs how much reconciliation can achieve, reliably only for homogeneous portfolios. We test this with data-driven intermediate hierarchies built through clustering: on the industrial portfolio, depth-3 and depth-4 hierarchies improve aggregate accuracy by up to 15.40% over flat baselines, across model families. On the more heterogeneous residential portfolio, the benefit depends on model family, improving accuracy for one and offering none or a net-loss for others. These results give three contributions: a quantification of the coherency gap across two real-world energy systems; a mechanistic account, through the Mean Absolute Adjustment metric, of how hierarchy depth redistributes reconciliation corrections across levels; and evidence that depth, not clustering-algorithm, drives the accuracy gain where it occurs, consistently across local and global training strategies.
- 22 August 2026
- Published by Elsevier in SSRN Electronic Journal
Open AccessNot Adding Up: Bridging Coherency Gaps in Energy Forecasts Using Hierarchical Clustering and Reconciliation
Accurate forecasts of distributed energy resource output are needed at multiple aggregation levels for grid operation and flexibility scheduling. Energy service companies forecast at the unit level while grid operators forecast at the feeder level, independently. Summing unit forecasts does not reproduce the feeder forecast, after accounting for losses. This inconsistency, the coherency gap, distorts estimates of demand or output and the flexibility needs that follow. We quantify this gap on two real-world datasets: residential-demand from 2,072 UK households and industrial combined heat and power generation from 82 Belgian units. Forecasts disagree by 3.5% to 8.8% of total load on average, growing with unit-level heterogeneity. Hierarchical reconciliation enforces aggregation consistency, but in shallow two-level hierarchies it yields only marginal accuracy gains, aggregation structure, not reconciliation methodology, governs how much reconciliation can achieve, reliably only for homogeneous portfolios. We test this with data-driven intermediate hierarchies built through clustering: on the industrial portfolio, depth-3 and depth-4 hierarchies improve aggregate accuracy by up to 15.40% over flat baselines, across model families. On the more heterogeneous residential portfolio, the benefit depends on model family, improving accuracy for one and offering none or a net-loss for others. These results give three contributions: a quantification of the coherency gap across two real-world energy systems; a mechanistic account, through the Mean Absolute Adjustment metric, of how hierarchy depth redistributes reconciliation corrections across levels; and evidence that depth, not clustering-algorithm, drives the accuracy gain where it occurs, consistently across local and global training strategies.
- 6 May 2026
- Published by Institute of Electrical and Electronics Engineers (IEEE)
- p. 1123-1127
Sustainable Building Energy Performance Assessment Using a Hybrid LLM-RAG Framework
Residential buildings account for a substantial share of electricity consumption, yet energy performance evaluation often requires technical expertise to interpret raw meter data. This study presents a hybrid large language model (LLM) and retrieval-augmented generation (RAG) framework for sustainable assessment of residential building energy performance. The proposed framework integrates structured time-series data and descriptive metadata with contextual knowledge retrieval to enable natural language (NL) interaction and generate interpretable textual responses and graphical visualisations. A six-and-a-half-month dataset from November 2018 to May 2019 was analysed to evaluate total and monthly energy consumption, peak demand characteristics, load factor, appliance contribution, daily load behavior, and seasonal transition effects. Results indicate a steady increase in coolingrelated demand toward early summer, with air-conditioning (AC) loads significantly influencing peak growth. Beyond numerical analysis, the framework translates computed metrics into interpretable textual summaries and visualisations, facilitating accessibility for non-specialist users. The system demonstrates how hybrid language model architectures can transform structured building energy data into actionable, usercentered sustainability insights.
- 6 May 2026
- Published by Institute of Electrical and Electronics Engineers (IEEE)
- p. 1123-1127
Sustainable Building Energy Performance Assessment Using a Hybrid LLM-RAG Framework
Residential buildings account for a substantial share of electricity consumption, yet energy performance evaluation often requires technical expertise to interpret raw meter data. This study presents a hybrid large language model (LLM) and retrieval-augmented generation (RAG) framework for sustainable assessment of residential building energy performance. The proposed framework integrates structured time-series data and descriptive metadata with contextual knowledge retrieval to enable natural language (NL) interaction and generate interpretable textual responses and graphical visualisations. A six-and-a-half-month dataset from November 2018 to May 2019 was analysed to evaluate total and monthly energy consumption, peak demand characteristics, load factor, appliance contribution, daily load behavior, and seasonal transition effects. Results indicate a steady increase in coolingrelated demand toward early summer, with air-conditioning (AC) loads significantly influencing peak growth. Beyond numerical analysis, the framework translates computed metrics into interpretable textual summaries and visualisations, facilitating accessibility for non-specialist users. The system demonstrates how hybrid language model architectures can transform structured building energy data into actionable, usercentered sustainability insights.
- 3 October 2025
- Published by Wiley in Business Strategy and the Environment
- Vol. 35 (2) , 1565-1585
Leveraging Green Intellectual Capital for Sustainability: The Strategic Role of Green Business Strategy and Green Technology Adoption
Sustainability research has yet to clarify how firms convert their environmental knowledge into measurable performance outcomes. Addressing this gap, we develop and test a novel dual‐pathway framework in which green intellectual capital (GIC), the firm's environmental know‐how and capabilities, drives sustainable performance through two intertwined mechanisms. Analyzing survey data from 367 manufacturing firms, our findings show that GIC significantly enhances sustainable performance when strategically embedded via a green business strategy and that the high levels of green technology adoption further strengthen the strategic conversion of GIC into environmental and economic gains. Our post hoc analysis further identifies distinct first‐order dimensions of GIC exert differential effects on sustainability, underscoring the need for targeted knowledge investments. The originality of this study lies in uncovering this moderated‐mediation mechanism for the first time, thereby integrating intellectual capital theory with resource‐based and dynamic‐capabilities perspectives in the sustainability domain. We advance understanding of how intangible environmental assets can be strategically and technologically leveraged, offering clarity on how firms can effectively harness their intangible assets for sustainability.
- 3 October 2025
- Published by Wiley in Business Strategy and the Environment
- Vol. 35 (2) , 1565-1585
Leveraging Green Intellectual Capital for Sustainability: The Strategic Role of Green Business Strategy and Green Technology Adoption
Sustainability research has yet to clarify how firms convert their environmental knowledge into measurable performance outcomes. Addressing this gap, we develop and test a novel dual‐pathway framework in which green intellectual capital (GIC), the firm's environmental know‐how and capabilities, drives sustainable performance through two intertwined mechanisms. Analyzing survey data from 367 manufacturing firms, our findings show that GIC significantly enhances sustainable performance when strategically embedded via a green business strategy and that the high levels of green technology adoption further strengthen the strategic conversion of GIC into environmental and economic gains. Our post hoc analysis further identifies distinct first‐order dimensions of GIC exert differential effects on sustainability, underscoring the need for targeted knowledge investments. The originality of this study lies in uncovering this moderated‐mediation mechanism for the first time, thereby integrating intellectual capital theory with resource‐based and dynamic‐capabilities perspectives in the sustainability domain. We advance understanding of how intangible environmental assets can be strategically and technologically leveraged, offering clarity on how firms can effectively harness their intangible assets for sustainability.
- 12 September 2024
- Published by Elsevier in International Journal of Production Economics
- Vol. 277
Dynamic pricing for perishable goods: A data-driven digital transformation approach
- 12 September 2024
- Published by Elsevier in International Journal of Production Economics
- Vol. 277
Dynamic pricing for perishable goods: A data-driven digital transformation approach
- Open Access
- 31 May 2024
- Published by Association for Computing Machinery (ACM)
- p. 692-695
Beyond empirical risk: testing data-driven energy models in sparse and biased data regimes
The energy system is undergoing a fundamental transition, with rapid proliferation of renewable energy sources and electrification of demand. This means that the status quo of carefully planning grid operation in a centralized manner is no longer tenable. Consequently, data-driven methods are increasingly being utilized to accurately forecast demand and generation, as well as to activate the necessary flexibility in case of errors in these forecasts. The often black-box nature of these methods means that their predictions cannot necessarily be relied upon. In energy, this has been historically addressed by a shift towards quantifying prediction uncertainty. In practice however, it is often difficult to produce truly sharp and well-calibrated uncertainty intervals, especially for out-of-sample situations. Moreover, quantifying uncertainty still does not address the inherent lack of interpretability and causality in these methods especially when trained on biased data. In this paper, we propose a novel post-hoc method, relying on domain-informed counterfactuals, to better understand the predictions of (black-box) predictive methods, and ascertain if they follow existing domain knowledge. We apply this to two different predictive problems in energy modelling, and show that the proposed method can provide insights into well-performing but un-trustworthy models.
- 31 May 2024
- Published by Association for Computing Machinery (ACM)
- p. 692-695
Open AccessBeyond empirical risk: testing data-driven energy models in sparse and biased data regimes
The energy system is undergoing a fundamental transition, with rapid proliferation of renewable energy sources and electrification of demand. This means that the status quo of carefully planning grid operation in a centralized manner is no longer tenable. Consequently, data-driven methods are increasingly being utilized to accurately forecast demand and generation, as well as to activate the necessary flexibility in case of errors in these forecasts. The often black-box nature of these methods means that their predictions cannot necessarily be relied upon. In energy, this has been historically addressed by a shift towards quantifying prediction uncertainty. In practice however, it is often difficult to produce truly sharp and well-calibrated uncertainty intervals, especially for out-of-sample situations. Moreover, quantifying uncertainty still does not address the inherent lack of interpretability and causality in these methods especially when trained on biased data. In this paper, we propose a novel post-hoc method, relying on domain-informed counterfactuals, to better understand the predictions of (black-box) predictive methods, and ascertain if they follow existing domain knowledge. We apply this to two different predictive problems in energy modelling, and show that the proposed method can provide insights into well-performing but un-trustworthy models.
- 3 June 2022
- Published by Elsevier in Energy Policy
- Vol. 167
The role of residential distributed energy resources in Pakistan's energy transition
- 3 June 2022
- Published by Elsevier in Energy Policy
- Vol. 167
The role of residential distributed energy resources in Pakistan's energy transition
- 17 September 2021
- Published by Taylor & Francis in European Journal of Information Systems
- Vol. 32 (3) , 462-484
Control-style ambidexterity and information systems project performance: an expanded view of control activities
Information systems (IS) projects represent key building blocks of large-scale digital transformation and innovation initiatives. As a result, IS projects have become increasingly ambitious in terms of both goals and scale, making it even more challenging for managers to exercise control over such projects. While prior research focused primarily on the direct and interactive effects of formal and informal control modes on IS project performance, recent research directs attention to the importance of considering project managers’ control styles (i.e., how managers interact with controlees to enact controls). Corresponding studies also indicate that “either/or” control approaches – as opposed to “both/and” approaches – are no longer viable in today’s complex environment. As such, our study draws on an ambidexterity perspective to theoretically develop and empirically test the direct and interactive effects of control-style ambidexterity on IS project performance. Using matched-pair data from 146 IS projects (from 146 high-tech firms), we find that control-style ambidexterity improves project performance – directly and in combination with both formal and informal control. The study results contribute novel insights regarding the effective control of IS projects in the digital era and help explain mixed findings in prior literature, thereby facilitating continued theory development in the research area.
- 17 September 2021
- Published by Taylor & Francis in European Journal of Information Systems
- Vol. 32 (3) , 462-484
Control-style ambidexterity and information systems project performance: an expanded view of control activities
Information systems (IS) projects represent key building blocks of large-scale digital transformation and innovation initiatives. As a result, IS projects have become increasingly ambitious in terms of both goals and scale, making it even more challenging for managers to exercise control over such projects. While prior research focused primarily on the direct and interactive effects of formal and informal control modes on IS project performance, recent research directs attention to the importance of considering project managers’ control styles (i.e., how managers interact with controlees to enact controls). Corresponding studies also indicate that “either/or” control approaches – as opposed to “both/and” approaches – are no longer viable in today’s complex environment. As such, our study draws on an ambidexterity perspective to theoretically develop and empirically test the direct and interactive effects of control-style ambidexterity on IS project performance. Using matched-pair data from 146 IS projects (from 146 high-tech firms), we find that control-style ambidexterity improves project performance – directly and in combination with both formal and informal control. The study results contribute novel insights regarding the effective control of IS projects in the digital era and help explain mixed findings in prior literature, thereby facilitating continued theory development in the research area.
- 1 September 2021
- Published by Elsevier in Renewable and Sustainable Energy Reviews
- Vol. 148
Towards data-driven energy communities: A review of open-source datasets, models and tools
- 1 September 2021
- Published by Elsevier in Renewable and Sustainable Energy Reviews
- Vol. 148
Towards data-driven energy communities: A review of open-source datasets, models and tools