Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

Forte: An Interactive Visual Analytic Tool for Trust-Augmented Net Load Forecasting

Kaustav Bhattacharjee, Soumya Kundu, Indrasis Chakraborty, Aritra Dasgupta

TL;DR

Forte addresses the challenge of trusting probabilistic net load forecasts by enabling post-hoc evaluation across input variables and simulated noise. The approach combines coordinated views (Net Load View, Inputs View, Experiment Design) with interactive controls to compare actual vs predicted nets and to quantify uncertainty with a $95\%$ CI. Key contributions include a design study with energy scientists, an integrated visualization workflow, and demonstrative scenarios showing how noise and data quality affect forecast performance, evidenced by metrics such as $MAE$ and $MAPE$. The work has practical impact for grid operators and policymakers seeking data-driven, trust-aware forecasting and decision support.

Abstract

Accurate net load forecasting is vital for energy planning, aiding decisions on trade and load distribution. However, assessing the performance of forecasting models across diverse input variables, like temperature and humidity, remains challenging, particularly for eliciting a high degree of trust in the model outcomes. In this context, there is a growing need for data-driven technological interventions to aid scientists in comprehending how models react to both noisy and clean input variables, thus shedding light on complex behaviors and fostering confidence in the outcomes. In this paper, we present Forte, a visual analytics-based application to explore deep probabilistic net load forecasting models across various input variables and understand the error rates for different scenarios. With carefully designed visual interventions, this web-based interface empowers scientists to derive insights about model performance by simulating diverse scenarios, facilitating an informed decision-making process. We discuss observations made using Forte and demonstrate the effectiveness of visualization techniques to provide valuable insights into the correlation between weather inputs and net load forecasts, ultimately advancing grid capabilities by improving trust in forecasting models.

Forte: An Interactive Visual Analytic Tool for Trust-Augmented Net Load Forecasting

TL;DR

Forte addresses the challenge of trusting probabilistic net load forecasts by enabling post-hoc evaluation across input variables and simulated noise. The approach combines coordinated views (Net Load View, Inputs View, Experiment Design) with interactive controls to compare actual vs predicted nets and to quantify uncertainty with a CI. Key contributions include a design study with energy scientists, an integrated visualization workflow, and demonstrative scenarios showing how noise and data quality affect forecast performance, evidenced by metrics such as and . The work has practical impact for grid operators and policymakers seeking data-driven, trust-aware forecasting and decision support.

Abstract

Accurate net load forecasting is vital for energy planning, aiding decisions on trade and load distribution. However, assessing the performance of forecasting models across diverse input variables, like temperature and humidity, remains challenging, particularly for eliciting a high degree of trust in the model outcomes. In this context, there is a growing need for data-driven technological interventions to aid scientists in comprehending how models react to both noisy and clean input variables, thus shedding light on complex behaviors and fostering confidence in the outcomes. In this paper, we present Forte, a visual analytics-based application to explore deep probabilistic net load forecasting models across various input variables and understand the error rates for different scenarios. With carefully designed visual interventions, this web-based interface empowers scientists to derive insights about model performance by simulating diverse scenarios, facilitating an informed decision-making process. We discuss observations made using Forte and demonstrate the effectiveness of visualization techniques to provide valuable insights into the correlation between weather inputs and net load forecasts, ultimately advancing grid capabilities by improving trust in forecasting models.
Paper Structure (6 sections, 2 figures)

This paper contains 6 sections, 2 figures.

Table of Contents

  1. Introduction
  2. Visual analytics-based design
  3. Goal: Understand Net Load Forecasts w.r.t Input Variables
  4. Goal: Compare model performance w.r.t Noisy Inputs
  5. Experimental Results
  6. Conclusion

Figures (2)

  • Figure 1: The interface for our net load forecasting visual analytic tool (Forte): (a) Our application facilitates the comparison of actual and predicted net load within the selected time frame and solar penetration levels as defined (b) through the Options Selection Area. Further, (c) the influence of various weather conditions on predictions can be explored via the Inputs View Area. The highlighted region shows instances of missing temperature data and resultant disagreement between predicted and actual net load within the same time period. These insights are valuable to grid operators as it allows them to review the data quality, evaluate its impact on model performance, and make recommendations for sensor/metering upgrade.
  • Figure 2: Experimental Results: (a) Our application Forte enables the design of experiments through the creation of noisy inputs using various factors; and the results (error rates) can be cross-compared across various months for both the input variables of (b, c) temperature and (d, e) humidity; (f) with the option to view detailed observations for each month. These insights generated through Forte are valuable to the user (a grid operator) to not only reveal the underlying dependence of the model outcome (net load prediction) on different input weather conditions but also better prepare ahead of any impending weather events (e.g., heat/cold wave).