pith. sign in

arxiv: 1906.08910 · v1 · pith:AAYNQ7OEnew · submitted 2019-06-21 · 💻 cs.CY

Zero Latency for Emergencies: A Machine Learning based Approach to Quantify Impact of Construction Projects on Emergency Response in Urban Settings

Zhengbo Zou , Semiha Ergan This is my paper

Pith reviewed 2026-05-25 19:08 UTC · model grok-4.3

classification 💻 cs.CY
keywords construction permitsemergency response timesmachine learningclusteringurban zonesNew York Citysupervised learning
0
0 comments X

The pith

Zones with similar construction permit distributions can be clustered so that supervised learning predicts their average emergency response times.

A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.

The paper defines a signature for each urban zone as the distribution of construction work types permitted there over time. Zones sharing similar signatures are grouped through clustering. Supervised learning models are then applied to historical emergency response records to forecast average response times within each group. The method was tested on New York City permit and response data and is presented as replicable for other cities that publish comparable records. A reader would care because construction projects alter street conditions that affect first-responder arrival, and the work offers a data-driven way to anticipate those effects without relying solely on individual case studies.

Core claim

Zones that share similar construction-activity signatures, defined by the distribution of permitted work types, can be clustered, after which supervised learning algorithms predict the average emergency response times for each cluster using historical records.

What carries the argument

The construction-activity signature of a zone, defined as the distribution of historical construction work types permitted in that zone over time, which supports clustering and subsequent response-time prediction.

Load-bearing premise

Similarity in the distribution of permitted construction work types across zones is a sufficient proxy for similarity in the actual traffic, road-closure, or other physical effects that alter emergency response times.

What would settle it

Direct comparison of zones with nearly identical permit-type distributions that nevertheless show materially different changes in measured emergency response times would undermine the clustering and prediction steps.

Figures

Figures reproduced from arXiv: 1906.08910 by Semiha Ergan, Zhengbo Zou.

Figure 1
Figure 1. Figure 1: Construction Work Types and Work Subtypes included in DOB Permit Data From work types and work subtypes, we can categorize the construction activities based on the possible effect they have on the emergency response times (i.e., interior works have minimal influence on the emergency response times because they are self-contained). Among the eight work types, construction equipment (including dumping chute,… view at source ↗
read the original abstract

Continuous construction and rehabilitation in urban settings have unavoidable impacts on arrival times of first responders to emergency locations. Current research efforts on emergency response assessments focus on case studies, where specific periods (e.g., super storm Sandy) of emergency response times are analyzed. Simulation based studies that aim to evaluate response times in relation to various constraints/fleet sizes also exist. However, they do not analyze how specific changes (e.g., new and ongoing construction projects) in urban settings impact emergency response times of first responders. This paper aims to fill the gap and proposes a novel approach to predict the expected emergency response time for a given location using the fabric of zones regarding construction activities. This approach relies on historical records of emergency response and construction permits issued by city agencies. The approach first defines the signature of a zone (by zip codes) for construction activities based on the distribution of historical construction work types permitted in that zone over time. Then, zones that share similar signatures are clustered to find if there exists a relationship between construction signatures and emergency response times. Next, supervised learning algorithms are deployed to predict the average emergency response times for each cluster. The approach was tested using New York City's construction permit and emergency response records, and can be easily replicated for other cities with similar public datasets. This study serves as the first step towards quantitatively understanding construction projects' impact on a quality of life (QoL) indicator (specifically emergency response times) in urban settings.

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, simulated authors' rebuttal, and a circularity audit. Tearing a paper down is the easy half of reading it; the pith above is the substance, this is the friction.

Referee Report

3 major / 1 minor

Summary. The manuscript proposes a machine learning pipeline to quantify construction impacts on emergency response times: it defines a 'zone signature' for each NYC zip code from the empirical distribution of permitted construction work types, clusters zones sharing similar signatures, and trains supervised models to predict average emergency response time from cluster membership. The approach is tested on historical NYC construction permit and emergency response records and presented as replicable for other cities.

Significance. If the clustering and prediction steps demonstrably isolate construction-induced effects rather than correlated zone traits, the work would supply a replicable, data-driven method for assessing a quality-of-life impact using only public permit and response records. The absence of any reported performance numbers, cross-validation, or baseline comparisons, however, leaves the predictive claim untested in the current manuscript.

major comments (3)
  1. [Abstract] Abstract (paragraph describing the signature definition and clustering step): the zone signature is constructed solely from the distribution of permit work types; no controls, matching, or regression adjustment for confounders (population density, land-use category, baseline road-network density) are described, so any observed cluster-to-response-time relationship may be driven by those co-varying factors rather than construction activity itself.
  2. [Abstract] Abstract (supervised learning paragraph): no performance metrics, cross-validation procedure, baseline comparisons (e.g., against a model using only zone-level covariates), or error analysis are supplied, rendering it impossible to determine whether cluster membership adds predictive value beyond chance or simple zone attributes.
  3. [Abstract] Abstract (testing sentence): the claim that the method 'quantifies' construction impact rests on the untested assumption that similarity in permit-type distributions is a sufficient proxy for similarity in physical disruptions (road closures, traffic impedance); the manuscript provides no falsification test or sensitivity analysis for this assumption.
minor comments (1)
  1. [Abstract] The abstract refers to 'the fabric of zones regarding construction activities' without defining the term; a brief clarification of this phrasing would improve readability.

Simulated Author's Rebuttal

3 responses · 0 unresolved

We thank the referee for the constructive feedback. We address each major comment below and indicate planned revisions to the manuscript.

read point-by-point responses

  1. Referee: [Abstract] Abstract (paragraph describing the signature definition and clustering step): the zone signature is constructed solely from the distribution of permit work types; no controls, matching, or regression adjustment for confounders (population density, land-use category, baseline road-network density) are described, so any observed cluster-to-response-time relationship may be driven by those co-varying factors rather than construction activity itself.

    Authors: We agree that the signature definition focuses exclusively on permit-type distributions without explicit controls for zone-level confounders. This design choice isolates construction activity patterns from available public data, but we recognize the risk of confounding. In revision we will add an explicit discussion of potential confounders (population density, land-use, road network) in the methods and limitations sections and clarify that the reported relationships are associative rather than causal. We will also note that incorporating additional covariates is a natural extension for future work. revision: partial

  2. Referee: [Abstract] Abstract (supervised learning paragraph): no performance metrics, cross-validation procedure, baseline comparisons (e.g., against a model using only zone-level covariates), or error analysis are supplied, rendering it impossible to determine whether cluster membership adds predictive value beyond chance or simple zone attributes.

    Authors: The full manuscript describes the supervised models but does not report quantitative performance numbers, cross-validation details, or baseline comparisons. We accept this omission and will revise the results section to include cross-validated metrics (e.g., RMSE, R^{2}), a baseline model using only zone covariates, and basic error analysis so that the incremental value of cluster membership can be evaluated. revision: yes

  3. Referee: [Abstract] Abstract (testing sentence): the claim that the method 'quantifies' construction impact rests on the untested assumption that similarity in permit-type distributions is a sufficient proxy for similarity in physical disruptions (road closures, traffic impedance); the manuscript provides no falsification test or sensitivity analysis for this assumption.

    Authors: The proxy assumption is necessary given the public data sources, which lack direct disruption measurements. The NYC historical records provide an empirical check of the overall pipeline. We will add a dedicated limitations paragraph acknowledging the assumption and will include a sensitivity analysis (varying the number of clusters and work-type granularity) in the revised manuscript to test robustness. revision: yes

Circularity Check

0 steps flagged

No circularity: standard clustering-plus-prediction pipeline on external records

full rationale

The paper extracts zone signatures directly from the empirical distribution of construction permit types, clusters zones on those signatures, and trains supervised models to predict average emergency response times from cluster membership or signatures. This workflow uses independent historical city datasets for both features and target variable; no equation, definition, or self-citation reduces the output prediction to a fitted parameter or input by construction. The derivation chain remains self-contained against external benchmarks.

Axiom & Free-Parameter Ledger

0 free parameters · 1 axioms · 1 invented entities

The approach rests on standard unsupervised and supervised learning routines plus the domain assumption that permit-type distributions capture the relevant construction effects.

axioms (1)
  • domain assumption Clustering on construction-permit distributions groups zones that experience comparable impacts on emergency response
    Invoked when the paper states that zones sharing signatures are clustered to find a relationship with response times.
invented entities (1)
  • Zone signature no independent evidence
    purpose: A vector representing the historical distribution of construction work types in a zip-code zone
    New representational construct introduced to enable the clustering step; no independent evidence supplied.

pith-pipeline@v0.9.0 · 5798 in / 1162 out tokens · 21178 ms · 2026-05-25T19:08:10.281494+00:00 · methodology

discussion (0)

Sign in with ORCID, Apple, or X to comment. Anyone can read and Pith papers without signing in.

Reference graph

Works this paper leans on

8 extracted references · 8 canonical work pages

  1. [1]

    Current research efforts on emergency response assessments focus on case studies, where specific periods (e.g., 9/11 and storm Sandy) of emergency response times are analyzed

    CSCE Annual Conference Growing with youth – Croître avec les jeunes Laval (Greater Montreal) June 12 - 15, 2019 ZERO LATENCY FOR EMERGENCIES: A MACHINE LEARNING BASED APPROACH TO QUANTIFY IMPACT OF CONSTRUCTION PROJECTS ON EMERGENCY RESPONSE IN URBAN SETTINGS Zhengbo Zou1 and Semiha Ergan2 1PhD Candidate, Department of Civil and Urban Engineering, New Yor...

    work page 2019

  2. [2]

    The analysis was kept in 2013-2017, where both datasets were available. We consider the contribution of this paper essential because: (1) this paper is the first known attempt to quantify building construction projects’ impact on the emergency response time on an urban scale, regardless of type of construction activities (e.g., foundation, new buildings, ...

    work page 2013

  3. [3]

    more than one million buildings and active construction sites in New York City by enforcing construction laws, building codes, and zoning resolutions

    DOB regulates “more than one million buildings and active construction sites in New York City by enforcing construction laws, building codes, and zoning resolutions” (DOB 2018). FDNY responds to fire and medical emergencies in the five boroughs of New York City. These datasets were used to evaluate the clustering and prediction performance of the approach...

    work page 2018

  4. [4]

    Construction Work Types and Work Subtypes included in DOB Permit Data From work types and work subtypes, we can categorize the construction activities based on the possible effect they have on the emergency response times (i.e., interior works have minimal influence on the emergency response times because they are self-contained). Among the eight work typ...

    work page 2013

  5. [5]

    Incident response time is defined as the time elapsed (in seconds) for the first responding unit to arrive at the incident location

    The columns used in the data analysis include two categories of information, namely, (1) location information (i.e., incident borough and zip code) and (2) time information (i.e., incident date, time, and incident response time). Incident response time is defined as the time elapsed (in seconds) for the first responding unit to arrive at the incident loca...

    work page 2016

  6. [6]

    Cluster 5 resembles to Cluster 4 in its construction signature, except for the new construction work (being lower), which is expected to change the response times in these regions

    Cluster 1 has the shortest response time, which is expected given it has the smallest alteration, foundation and new buildings works in its distribution. Cluster 5 resembles to Cluster 4 in its construction signature, except for the new construction work (being lower), which is expected to change the response times in these regions. Response time of Clust...

    work page 2017

  7. [7]

    OR/MS research in disaster operations management

    Cluster 3 was again excluded for its small sample size. Table 3: Regression Model Performance Measured by R Squared for Average Emergency Response Time Cluster Ordinary Least Squares Decision Tree Random Forest 1 0.59 0.72 0.76 2 0.50 0.67 0.77 4 0.52 0.79 0.81 5 0.54 0.65 0.72 One observation from Table 3 is that Random Forest showed the highest R-Square...

    work page 2006

  8. [8]

    Selecting 3D urban visualisation models for disaster management: a rule-based approach

    Wildfires in Southern California.” Journal of Forestry, 102(7), 26-31. Kemec, S., Zlatanova, S., & Duzgun, S. (2009, June). “Selecting 3D urban visualisation models for disaster management: a rule-based approach”. In Proceedings of TIEMS 2009 Annual Conference, June (pp. 9-11). Kolesar, P., & Blum, E. H. (1973). “Square root laws for fire engine response ...

    work page arXiv 2009