Kun Liu, S. Guhathakurta, Chaeyeon Han, E. Hittinger, Sinoun Phoung, Eric Williams
As office workers shift to telework, office building space requirements should decrease, but this relationship has not been empirically studied. We construct a dataset describing historical office building space, number of office workers, and number of teleworkers from 2003-2019 in the US, and use linear regression to estimate the effect of telework on office building space. The results show that the average office building space required for an additional office worker and teleworker is 32 and 18 square meters (340 and 191 square feet), respectively, suggesting an average 44% reduction in office building space when an office worker transitions to telework.
{"title":"How much is US Office Building Space Reduced per Teleworker?","authors":"Kun Liu, S. Guhathakurta, Chaeyeon Han, E. Hittinger, Sinoun Phoung, Eric Williams","doi":"10.32866/001c.115400","DOIUrl":"https://doi.org/10.32866/001c.115400","url":null,"abstract":"As office workers shift to telework, office building space requirements should decrease, but this relationship has not been empirically studied. We construct a dataset describing historical office building space, number of office workers, and number of teleworkers from 2003-2019 in the US, and use linear regression to estimate the effect of telework on office building space. The results show that the average office building space required for an additional office worker and teleworker is 32 and 18 square meters (340 and 191 square feet), respectively, suggesting an average 44% reduction in office building space when an office worker transitions to telework.","PeriodicalId":508951,"journal":{"name":"Findings","volume":"35 26","pages":""},"PeriodicalIF":0.0,"publicationDate":"2024-04-02","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"140753133","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Time use statistics are widely documented in order to track societal trends. The most reliable and widely used collection method for time use data are diaries in which a person documents all activities participated in for a given period. This study uncovers daily time use patterns across the entire adult population of German-speaking Switzerland with such data. The findings are the first diary-based account of time use for the region and illuminate important differences between genders and based on parenthood status in terms of paid and unpaid work, as well as working from home.
{"title":"How do the Swiss Spend their Time?","authors":"Caroline Winkler, K. Axhausen","doi":"10.32866/001c.108600","DOIUrl":"https://doi.org/10.32866/001c.108600","url":null,"abstract":"Time use statistics are widely documented in order to track societal trends. The most reliable and widely used collection method for time use data are diaries in which a person documents all activities participated in for a given period. This study uncovers daily time use patterns across the entire adult population of German-speaking Switzerland with such data. The findings are the first diary-based account of time use for the region and illuminate important differences between genders and based on parenthood status in terms of paid and unpaid work, as well as working from home.","PeriodicalId":508951,"journal":{"name":"Findings","volume":"58 15","pages":""},"PeriodicalIF":0.0,"publicationDate":"2024-03-27","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"140376349","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Emil G. Dimanchev, Stein-Erik Fleten, Steven A Gabriel, Magnus Korpas
Analyses of climate policies often assume the economy is in a first-best equilibrium with well-functioning markets. This paper studies policy effects in power systems characterized by a market failure known as the missing market problem, whereby the incompleteness of long-term markets leaves investors exposed to uninsured risk. We find that renewable tax credits and CO 2 taxes may partly correct this market failure, thus providing an economic benefit additional to climate change mitigation. Consequently, illustrative experiments show the costs of these policies to be lower, and in some cases even negative, in power systems with missing risk markets.
{"title":"Effects of Electricity Sector Climate Policies in a Second-best World of Missing Risk Markets","authors":"Emil G. Dimanchev, Stein-Erik Fleten, Steven A Gabriel, Magnus Korpas","doi":"10.32866/001c.94993","DOIUrl":"https://doi.org/10.32866/001c.94993","url":null,"abstract":"Analyses of climate policies often assume the economy is in a first-best equilibrium with well-functioning markets. This paper studies policy effects in power systems characterized by a market failure known as the missing market problem, whereby the incompleteness of long-term markets leaves investors exposed to uninsured risk. We find that renewable tax credits and CO 2 taxes may partly correct this market failure, thus providing an economic benefit additional to climate change mitigation. Consequently, illustrative experiments show the costs of these policies to be lower, and in some cases even negative, in power systems with missing risk markets.","PeriodicalId":508951,"journal":{"name":"Findings","volume":"41 1","pages":""},"PeriodicalIF":0.0,"publicationDate":"2024-03-19","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"140229325","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Frédérick Chabot, Nicolas Saunier, Geneviève Boisjoly
There is growing interest in the study of the use of streets and adjacent public spaces. This paper highlights a case study conducted at a public square in Montreal. The objective is to evaluate a mobile data collection tool developed for the study of public spaces and its performance, while characterizing the use of the space. To do so, five observers made over 1200 unique observations. The analysis shows the variability between observers differs according to the observed variable and nearly 4 % of the clicks were made to correct an erroneous input.
{"title":"Testing a New Mobile Application to Study Public Spaces","authors":"Frédérick Chabot, Nicolas Saunier, Geneviève Boisjoly","doi":"10.32866/001c.94712","DOIUrl":"https://doi.org/10.32866/001c.94712","url":null,"abstract":"There is growing interest in the study of the use of streets and adjacent public spaces. This paper highlights a case study conducted at a public square in Montreal. The objective is to evaluate a mobile data collection tool developed for the study of public spaces and its performance, while characterizing the use of the space. To do so, five observers made over 1200 unique observations. The analysis shows the variability between observers differs according to the observed variable and nearly 4 % of the clicks were made to correct an erroneous input.","PeriodicalId":508951,"journal":{"name":"Findings","volume":"21 2","pages":""},"PeriodicalIF":0.0,"publicationDate":"2024-03-13","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"140247922","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
This study delves into mismatches between accessibility indicators and perceived accessibility across transport modes for the case of grocery shopping. Conducted in Gothenburg, Sweden, the study combines a web panel survey with 1,423 participants and detailed location-based accessibility indicators. Findings reveal mismatches, with analyst’s overestimation (when the accessibility indicator is high, despite low perceived access) and analyst’s underestimation (low indicator, high perceived accessibility) varying across transportation modes. Notably, underestimation is prominent for car accessibility. Multinomial logistic regressions identify key variables influencing these mismatches, such as parenting status, education level and habitual car use.
{"title":"When is Perceived Accessibility Over- or Underestimated by Accessibility Indicators?","authors":"Evangelos Vafeiadis, Erik Elldér","doi":"10.32866/001c.94648","DOIUrl":"https://doi.org/10.32866/001c.94648","url":null,"abstract":"This study delves into mismatches between accessibility indicators and perceived accessibility across transport modes for the case of grocery shopping. Conducted in Gothenburg, Sweden, the study combines a web panel survey with 1,423 participants and detailed location-based accessibility indicators. Findings reveal mismatches, with analyst’s overestimation (when the accessibility indicator is high, despite low perceived access) and analyst’s underestimation (low indicator, high perceived accessibility) varying across transportation modes. Notably, underestimation is prominent for car accessibility. Multinomial logistic regressions identify key variables influencing these mismatches, such as parenting status, education level and habitual car use.","PeriodicalId":508951,"journal":{"name":"Findings","volume":"29 7","pages":""},"PeriodicalIF":0.0,"publicationDate":"2024-03-12","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"140248871","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Artificial intelligence (AI) tools (in particular Large Language Models) have the potential to reduce the time needed to perform thematic analysis. To better understand their potential in the transportation field, we compare human-based to AI-based outcomes. Our findings indicate that AI tools, such as ChatGPT, could synthetize and summarize major topics present in our dataset regardless of previous user exposure to the subject or not. Nonetheless, caution is required as results might miss the nuance of less frequent themes. These tools could be used to accelerate the process under the supervision of researchers and practitioners given responder consent and the following of ethical practices.
{"title":"A Comparison of the Results from Artificial Intelligence-based and Human-based Transport-related Thematic Analysis","authors":"Thiago Carvalho, Hisham Negm, A. El-geneidy","doi":"10.32866/001c.94401","DOIUrl":"https://doi.org/10.32866/001c.94401","url":null,"abstract":"Artificial intelligence (AI) tools (in particular Large Language Models) have the potential to reduce the time needed to perform thematic analysis. To better understand their potential in the transportation field, we compare human-based to AI-based outcomes. Our findings indicate that AI tools, such as ChatGPT, could synthetize and summarize major topics present in our dataset regardless of previous user exposure to the subject or not. Nonetheless, caution is required as results might miss the nuance of less frequent themes. These tools could be used to accelerate the process under the supervision of researchers and practitioners given responder consent and the following of ethical practices.","PeriodicalId":508951,"journal":{"name":"Findings","volume":"38 4","pages":""},"PeriodicalIF":0.0,"publicationDate":"2024-03-07","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"140258434","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Neighborhood attributes may have different relationships with neighborhood satisfaction and life satisfaction. However, few studies have compared linkages between the neighborhood environment and these two satisfaction outcomes. This study fills this gap by illustrating how the same neighborhood attributes relate to neighborhood satisfaction and life satisfaction differently. Results indicate that the neighborhood environment has a stronger association with neighborhood satisfaction than with life satisfaction. Moreover, attributes pertinent to neighborhood appearance and basic features have the strongest connection with neighborhood satisfaction, whereas attributes regarding leisure activities and social cohesion show more substantial correlations with life satisfaction than other neighborhood attributes.
{"title":"How Do Neighborhood Attributes Relate to Life Satisfaction and Neighborhood Satisfaction Differently?","authors":"Xinyi Wu, Jason Cao, Yingling Fan, A. Ramaswami","doi":"10.32866/001c.92769","DOIUrl":"https://doi.org/10.32866/001c.92769","url":null,"abstract":"Neighborhood attributes may have different relationships with neighborhood satisfaction and life satisfaction. However, few studies have compared linkages between the neighborhood environment and these two satisfaction outcomes. This study fills this gap by illustrating how the same neighborhood attributes relate to neighborhood satisfaction and life satisfaction differently. Results indicate that the neighborhood environment has a stronger association with neighborhood satisfaction than with life satisfaction. Moreover, attributes pertinent to neighborhood appearance and basic features have the strongest connection with neighborhood satisfaction, whereas attributes regarding leisure activities and social cohesion show more substantial correlations with life satisfaction than other neighborhood attributes.","PeriodicalId":508951,"journal":{"name":"Findings","volume":"85 16","pages":""},"PeriodicalIF":0.0,"publicationDate":"2024-03-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"140086614","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Parastoo Jabbari, A. Ranjbari, P. Leiby, Don Mackenzie
How does the value of travel time (VOTT) differ between time spent driving and time spent being driven in a car? We examined revealed choices between ridehailing and free-float carsharing using data from an aggregator app that allowed users to choose between these alternatives based on real-time conditions. We used a mixed logit model to control for price, in-vehicle time, and out-of-vehicle time (walk or wait time). The model results indicate that VOTT declines by an average of $23 per hour (approximately 60%) for members of our sample when riding in a ridehailing vehicle versus driving a carsharing vehicle.
{"title":"Value of Travel Time is Lower When Being Driven than when Driving Oneself","authors":"Parastoo Jabbari, A. Ranjbari, P. Leiby, Don Mackenzie","doi":"10.32866/001c.93915","DOIUrl":"https://doi.org/10.32866/001c.93915","url":null,"abstract":"How does the value of travel time (VOTT) differ between time spent driving and time spent being driven in a car? We examined revealed choices between ridehailing and free-float carsharing using data from an aggregator app that allowed users to choose between these alternatives based on real-time conditions. We used a mixed logit model to control for price, in-vehicle time, and out-of-vehicle time (walk or wait time). The model results indicate that VOTT declines by an average of $23 per hour (approximately 60%) for members of our sample when riding in a ridehailing vehicle versus driving a carsharing vehicle.","PeriodicalId":508951,"journal":{"name":"Findings","volume":"54 6","pages":""},"PeriodicalIF":0.0,"publicationDate":"2024-02-27","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"140426773","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2024-02-24DOI: 10.48550/arXiv.2402.15833
Yao Qiang, Subhrangshu Nandi, Ninareh Mehrabi, G. V. Steeg, Anoop Kumar, Anna Rumshisky, A. Galstyan
Large language models (LLMs) have demonstrated impressive performance on a number of natural language processing tasks, such as question answering and text summarization. However, their performance on sequence labeling tasks such as intent classification and slot filling (IC-SF), which is a central component in personal assistant systems, lags significantly behind discriminative models. Furthermore, there is a lack of substantive research on robustness of LLMs to various perturbations in the input prompts. The contributions of this paper are three-fold. First, we show that fine-tuning sufficiently large LLMs can produce IC-SF performance comparable to discriminative models. Next, we systematically analyze the performance deterioration of those fine-tuned models due to three distinct yet relevant types of input perturbations - oronyms, synonyms, and paraphrasing. Finally, we propose an efficient mitigation approach, Prompt Perturbation Consistency Learning (PPCL), which works by regularizing the divergence between losses from clean and perturbed samples. Our experiments show that PPCL can recover on an average 59% and 69% of the performance drop for IC and SF tasks, respectively. Furthermore, PPCL beats data augmentation approach while using ten times fewer augmented data samples.
大型语言模型（LLM）在许多自然语言处理任务（如问题解答和文本摘要）中都表现出了令人印象深刻的性能。然而，它们在序列标注任务（如个人助理系统的核心组件--意图分类和槽填充（IC-SF））上的表现却明显落后于判别模型。此外，关于 LLM 对输入提示中各种扰动的鲁棒性还缺乏实质性的研究。本文的贡献有三方面。首先，我们证明了微调足够大的 LLM 可以产生与判别模型相当的 IC-SF 性能。接下来，我们系统地分析了这些微调模型的性能因三种不同但相关的输入扰动类型--同义词、近义词和意译--而下降的情况。最后，我们提出了一种高效的缓解方法--即时扰动一致性学习（PPCL），该方法通过规范化来自干净样本和扰动样本的损失之间的差异来发挥作用。我们的实验表明，对于 IC 和 SF 任务，PPCL 平均可分别恢复 59% 和 69% 的性能下降。此外，PPCL 比数据增强方法少用十倍的增强数据样本。
{"title":"Prompt Perturbation Consistency Learning for Robust Language Models","authors":"Yao Qiang, Subhrangshu Nandi, Ninareh Mehrabi, G. V. Steeg, Anoop Kumar, Anna Rumshisky, A. Galstyan","doi":"10.48550/arXiv.2402.15833","DOIUrl":"https://doi.org/10.48550/arXiv.2402.15833","url":null,"abstract":"Large language models (LLMs) have demonstrated impressive performance on a number of natural language processing tasks, such as question answering and text summarization. However, their performance on sequence labeling tasks such as intent classification and slot filling (IC-SF), which is a central component in personal assistant systems, lags significantly behind discriminative models. Furthermore, there is a lack of substantive research on robustness of LLMs to various perturbations in the input prompts. The contributions of this paper are three-fold. First, we show that fine-tuning sufficiently large LLMs can produce IC-SF performance comparable to discriminative models. Next, we systematically analyze the performance deterioration of those fine-tuned models due to three distinct yet relevant types of input perturbations - oronyms, synonyms, and paraphrasing. Finally, we propose an efficient mitigation approach, Prompt Perturbation Consistency Learning (PPCL), which works by regularizing the divergence between losses from clean and perturbed samples. Our experiments show that PPCL can recover on an average 59% and 69% of the performance drop for IC and SF tasks, respectively. Furthermore, PPCL beats data augmentation approach while using ten times fewer augmented data samples.","PeriodicalId":508951,"journal":{"name":"Findings","volume":"54 4","pages":"1357-1370"},"PeriodicalIF":0.0,"publicationDate":"2024-02-24","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"140434257","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
This study analyzes visitor dynamics during the 2017 Total Solar Eclipse in Southern Illinois using data from X (formerly known as Twitter). Focusing on spatial and temporal patterns, we identified key visitor clusters and sentiments. The majority of visitors originated from Chicago, IL, Nashville, TN, and St. Louis, MO. Findings revealed concentrated activities in specific locations, with generally positive experiences shared on social media. Insights gained will aid in planning for the upcoming 2024 eclipse, enhancing visitor experiences and economic benefits for the region. This research underscores the value of social media data in understanding and managing large-scale events in rural areas.
{"title":"Preparing for 2024 Total Solar Eclipse: Mining Social Media Data to Understand Spectator Experience","authors":"Ruopu Li, Joseph Kalinzi","doi":"10.32866/001c.94197","DOIUrl":"https://doi.org/10.32866/001c.94197","url":null,"abstract":"This study analyzes visitor dynamics during the 2017 Total Solar Eclipse in Southern Illinois using data from X (formerly known as Twitter). Focusing on spatial and temporal patterns, we identified key visitor clusters and sentiments. The majority of visitors originated from Chicago, IL, Nashville, TN, and St. Louis, MO. Findings revealed concentrated activities in specific locations, with generally positive experiences shared on social media. Insights gained will aid in planning for the upcoming 2024 eclipse, enhancing visitor experiences and economic benefits for the region. This research underscores the value of social media data in understanding and managing large-scale events in rural areas.","PeriodicalId":508951,"journal":{"name":"Findings","volume":"7 3","pages":""},"PeriodicalIF":0.0,"publicationDate":"2024-02-23","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"140436881","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
京公网安备 11010802042870号
京ICP备2023020795号-1
扫码关注我们
求助内容：
应助结果提醒方式：