<?xml version="1.0" encoding="UTF-8"?><rss xmlns:dc="http://purl.org/dc/elements/1.1/" xmlns:content="http://purl.org/rss/1.0/modules/content/" xmlns:atom="http://www.w3.org/2005/Atom" version="2.0" xmlns:cc="http://cyber.law.harvard.edu/rss/creativeCommonsRssModule.html">
    <channel>
        <title><![CDATA[International Public Safety Data Institute - Medium]]></title>
        <description><![CDATA[Enhancing public safety through data science. - Medium]]></description>
        <link>https://medium.com/ipsdi?source=rss----ccb76b4bdcdd---4</link>
        <image>
            <url>https://cdn-images-1.medium.com/proxy/1*TGH72Nnw24QL3iV9IOm4VA.png</url>
            <title>International Public Safety Data Institute - Medium</title>
            <link>https://medium.com/ipsdi?source=rss----ccb76b4bdcdd---4</link>
        </image>
        <generator>Medium</generator>
        <lastBuildDate>Tue, 17 Nov 2026 06:33:47 GMT</lastBuildDate>
        <atom:link href="https://medium.com/feed/ipsdi" rel="self" type="application/rss+xml"/>
        <webMaster><![CDATA[yourfriends@medium.com]]></webMaster>
        <atom:link href="http://medium.superfeedr.com/" rel="hub"/>
        <item>
            <title><![CDATA[A simple, transparent move-up model in NFORS]]></title>
            <link>https://medium.com/ipsdi/a-simple-transparent-move-up-model-in-nfors-acb38cc07a1b?source=rss----ccb76b4bdcdd---4</link>
            <guid isPermaLink="false">https://medium.com/p/acb38cc07a1b</guid>
            <category><![CDATA[emergency-response]]></category>
            <category><![CDATA[gis]]></category>
            <category><![CDATA[fire-department]]></category>
            <dc:creator><![CDATA[Tyler Buffington, PhD]]></dc:creator>
            <pubDate>Mon, 03 Aug 2020 17:59:14 GMT</pubDate>
            <atom:updated>2020-08-04T02:34:10.077Z</atom:updated>
            <content:encoded><![CDATA[<h3>A simple, transparent move-up model in NFORS</h3><p>A common feature request from NFORS users is a tool that can inform decision makers of move-up strategies for heavy emergency call volume or unexpected surges in requests for service. Given the large amount of emergency incident data collected using NFORS, this functionality is possible. In this blog post, the methodology for an initial move-up model that leverages historical incident data and Geographic Information System (GIS) technology and provides recommendations that optimize a fire department’s resource coverage is described. The project goal is to produce a model that is simple and transparent enough that users can understand exactly how the model generates its results, but sophisticated enough to provide relevant and actionable information and recommendations.</p><h3>Why are move-ups necessary?</h3><p>Before we describe the methodology of the model, it is important to illustrate why temporarily transferring units can be advantageous in an emergency response system. First, let’s consider a basic incident heat map generated using the NFORS data from the Richmond Fire Department.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/640/0*OGugjXtfvTCJZ4Rb" /><figcaption><em>An incident density heat map.</em></figcaption></figure><p>In the figure above, the red regions indicate an area where incidents are most likely to occur. The blue regions indicate areas where incidents are least likely to occur. One of the ways to evaluate a department’s coverage capability is to determine what fraction of incidents occur within a specified travel time from at least one available emergency response unit. Travel time can vary, but as an example, let’s specify any location that is within an eight-minute travel time from an available emergency response unit. This area is assumed to be “covered” and any location that is more than an eight-minute travel time from an available emergency response unit is assumed to be “uncovered.” Using GIS technology, we can determine every location within a fire department’s jurisdictional boundary that is covered according to this definition. Then an incident density heat map can be plotted over the uncovered regions. The resulting map will provide a useful visualization of fire department coverage holes.</p><p>First, let’s consider the maximum coverage case, meaning that every station has an available unit. Note that these analyses are usually done for a specific unit type. This blog post focuses on engines as an example, but the same analyses can be done for other unit types (ALS, Rescue Unit, etc.). The covered area (the cross-hatched region) is computed by taking the union of the eight minute travel time polygons for every station. The result is shown below.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/640/0*tdukefn-E11ebC7g" /><figcaption><em>An example of a coverage visualization for Richmond Fire Rescue Department. The cross-hatched regions indicate “covered” locations that are within an 8-minute drive time from at least one station with an available unit. The “uncovered” regions are more than an eight minute drive time from a station, indicated in blue on the incident density heat map. This map represents the best possible coverage, when every station has at least one unit that is available and able to respond. Note, the uncovered region in the middle of the Richmond map is largely over a river.</em></figcaption></figure><p>As noted above, the previous visualization represents the case for which every station has an available unit. This situation is not always the case. When stations do not have an available emergency response unit to send to a call for help, they are excluded from the calculation of the covered area. Some incidents require multiple units, which can leave multiple stations without any emergency response units remaining to respond to new incidents. In this situation, units are deployed from other stations to a call for help or they may be repositioned in the vacant station to be available for another incident that may occur. For example, on April 13, 2017, <a href="https://wtvr.com/2017/04/13/crews-battling-second-alarm-fire-at-southside-warehouse/">a large fire</a> broke out in a former tobacco factory in Richmond. The department deployed fire engines from ten different stations to extinguish the fire and rescue any trapped occupants. In a situation like this, the ten fire stations are left unable to send these same fire engines to any subsequent new incidents (since they are already on a call). The resulting emergency response area coverage holes are illustrated below.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/640/0*E0KSQiAzxgp16Eza" /><figcaption><em>A visualization of the coverage holes that would have occurred without unit transfers during the tobacco factory fire of 2017. Ten stations dispatched their fire engines to the factory location. Without the transfer of neighboring emergency response units known as move-ups, this situation would have left busy regions of the city uncovered for a subsequent emergency call for help.</em></figcaption></figure><p>As shown above, some of the stations that sent engines to the tobacco factory fire are in areas of high call volume. However the original heat map shows that some of Richmond’s stations are in regions of relatively low call volume. Therefore, moving units from low call volume stations to high call volume stations during periods of high emergency call volume can help fill coverage holes. It is important to note that Richmond conducted emergency response unit transfers during this fire incident, so the figure above does not show the actual coverage profile during the incident. Nevertheless, this example illustrates why unit transfers can be beneficial, and the goal of our move-up model is to provide a quantitative framework for evaluating and recommending unit transfer strategies.</p><h3>How the model works</h3><p>At a high level, the model works by solving two optimization problems:</p><ol><li>If <em>n </em>units of a specific type (i.e. engine, rescue unit, etc.) are available, which <em>n </em>stations should house the available units to optimize coverage?</li><li>During times of high call volume, what is the best strategy for transferring less busy or available units from their current locations to these busy stations (areas)?</li></ol><h3>Determining ideal station locations</h3><p>The first step in finding the best station locations is to determine what “best” means. Although there are many appropriate figures of merit, the current version of the NFORS model attempts to maximize the fraction of incidents that occur within a specified drive time of at least one available unit. In order to show how this is done, let’s consider a simple example.</p><p>Imagine that a fire department consists of five stations. The coverage objective is to maximize the fraction of incidents that are within an eight-minute drivetime from at least one available unit in these stations. Using GIS software, a polygon (isochrone) that indicates every location that is within a eight-minute drivetime from each station can be obtained. If a location is within a station’s isochrone, it is “covered” by that station. To gain an idea of the call volume covered by each station, a random sample of 1,000 historical incident locations is taken. The assumption is that this sample gives insight into where new incidents are likely to occur. An illustration of the simplified example is shown below. Note that when showing a relatively small number of incidents, it is easier to visualize the individual incident locations as shown below. When showing a large number of incidents, a heat map is more useful.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/519/0*RO04-Jml8DYz23Ac" /><figcaption><em>A hypothetical department layout for illustrating the move-up model methodology. Five stations are indicated by the five dots. Surrounding each station is a hypothetical eight-minute drivetime polygon (isochrone). The 12 red “x” markers indicate the locations of past incidents in a random sample. Keep in mind that this is only an illustrative example. The actual model uses a random sample of 1,000 incidents and the drivetime polygons are typically more complicated than the circles shown.</em></figcaption></figure><p>Now, let’s say that only two emergency response units are available. This means that we have to pick the two stations that cover the most incidents. This is known as the <em>maximum coverage </em>problem, and a <a href="https://en.wikipedia.org/wiki/Greedy_algorithm"><em>greedy algorithm</em></a> is often used for these problems. This process means that we build the list of ideal stations by adding the busiest and then sequentially adding the next station that covers the most incidents that are not covered by a station already on the list. In our example, we can simply count how many incidents are covered by each of the five stations and rank them:</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/292/1*Az9jGmCVFFbEf3DLL8M0oA.png" /></figure><p>Because Station C covers more incidents than any of the other stations, it is the first “ideal” station added to the list. Now we repeat this process, excluding station C and all of the incidents it covers:</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/519/0*-af5oIbWG8tRZMVE" /></figure><p>Based on the above figure, the rankings are</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/307/1*YyYBS2bsrhU4DqF99fAl9Q.png" /></figure><p>Station D covers more incidents that have not already been covered by Station C, so it is added to the list of ideal stations. Therefore, the two “ideal” stations to house the two available units are Station C and Station D. As shown below, this results in an aggregated coverage polygon that contains 9 out of 12 (75%) of the incidents in the sample, which is more than any other set of two stations.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/519/0*5Oe5R992wAmxN17_" /></figure><p>Note that Station B technically covers more incidents in total than Station D. However, the coverage overlap between B and C makes it so that C and D form a better set than B and C (9 out of 12 incidents vs. 8 out of 12 incidents).</p><h3>Determining the movement strategy</h3><p>Now that the set of ideal stations has been identified, the next step is to determine the best strategy for moving units from their current locations to the ideal stations during times of high call volume in those areas. Specifically, the model identifies the set of moves that minimizes the total distance traveled for the move-up units. This is a special case of the <a href="https://en.wikipedia.org/wiki/Assignment_problem"><em>assignment problem</em></a><em>.</em></p><p>In our simplified example, let’s say that the two available units are originally at Stations A and B. This results in two possible movement strategies:</p><ol><li>Unit A moves to Station C; Unit B moves to Station D (the blue strategy below)</li><li>Unit A moves to Station D; Unit B moves to Station C (the red strategy below)</li></ol><p>These strategies are illustrated below:</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/519/0*GAFx45EernNCEMKe" /><figcaption><em>A visual illustration of the two movement strategies. Stations C and D have been identified as the ideal stations to house available units. However, the available units are originally at stations A and B. To move the available units to the ideal stations there are two possible strategies. The blue strategy is to move Unit A to Station C and move Unit B to Station D; the red strategy is to move Unit A to Station D and to move Unit B to Station C.</em></figcaption></figure><p>Now a cost must be assigned to each strategy. This assignment can be done via travel time, travel distance, or some other custom heuristic. For simplicity, the cost of a strategy will be estimated by total distance traveled. The distances for each of the potential moves are shown in the table below:</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/953/1*eMAZxBlvMxUm-3CKDPD7-A.png" /></figure><p>The total distance traveled for the blue strategy is 2+1=3 miles. The total movement for the red strategy is 1+1.4=2.4 miles. As a result, the red strategy would be recommended by the model since it minimizes the total distance traveled. In this example, this result is easy to see because there are only two possible movement strategies. In actual use cases, there are often many different potential strategies so it is more computationally efficient to use a technique known as the Hungarian algorithm to find the optimum strategy. The details of this algorithm are beyond the scope of this blog post, but are explained <a href="https://en.wikipedia.org/wiki/Hungarian_algorithm">here</a>.</p><h3>Limitations</h3><p>Although the current version of the move-up model can provide insight into the effectiveness of potential move-up strategies, it has several limitations. First of all, it always recommends the strategy that optimizes coverage, even if some of the moves provide small coverage benefits and require units to travel long distances. In future versions of the model, the user will be able to specify additional criteria so that the model will only recommend moves above a certain distance if the coverage benefit is above a certain threshold.</p><p>The model also optimizes for single unit coverage, meaning that if a location is within a specified travel time of an available unit, then it is assumed to be covered. In some cases, it may be advantageous to have more than just one unit covering locations with large call volumes. Furthermore, if one station has two available units, and another station has no available units, the model will always recommend moving one of the available units to the station with no available units. While the current methodology does not consider the potential impact of mutual aid, these effects may be added as departments request this feature.</p><h3>Conclusions</h3><p>This blog post outlines the methodology of the initial move-up model in NFORS. The goal of this model is to provide insight into the effectiveness of movement strategies using NFORS data. This initial version is relatively simple, but it provides a starting point for future iterations. Our goal is to continually improve the model based on user feedback. If you have any questions or comments, please contact us at <a href="/service/mailto:hello@i-psdi.org">hello@i-psdi.org</a>. If your department does not yet participate in NFORS, schedule a demo to see the power of analytics <a href="https://www.eventbrite.com/e/nfors-demo-registration-45782334194">https://www.eventbrite.com/e/nfors-demo-registration-45782334194</a>.</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=acb38cc07a1b" width="1" height="1" alt=""><hr><p><a href="https://medium.com/ipsdi/a-simple-transparent-move-up-model-in-nfors-acb38cc07a1b">A simple, transparent move-up model in NFORS</a> was originally published in <a href="https://medium.com/ipsdi">International Public Safety Data Institute</a> on Medium, where people are continuing the conversation by highlighting and responding to this story.</p>]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[Exploring the impact of social distancing on emergency call volume using Google’s mobility dataset]]></title>
            <link>https://medium.com/ipsdi/exploring-the-impact-of-social-distancing-on-emergency-call-volume-using-googles-mobility-dataset-a17f248f224c?source=rss----ccb76b4bdcdd---4</link>
            <guid isPermaLink="false">https://medium.com/p/a17f248f224c</guid>
            <category><![CDATA[fire-department]]></category>
            <category><![CDATA[socialdistancing]]></category>
            <category><![CDATA[emergency]]></category>
            <category><![CDATA[mobility]]></category>
            <category><![CDATA[traffic-accidents]]></category>
            <dc:creator><![CDATA[Tyler Buffington, PhD]]></dc:creator>
            <pubDate>Fri, 08 May 2020 17:19:37 GMT</pubDate>
            <atom:updated>2020-05-16T04:51:43.426Z</atom:updated>
            <content:encoded><![CDATA[<p>The COVID-19 pandemic has altered our lifestyles in an unprecedented way. In the last two months, there has been a dramatic reduction in the number of people commuting to work, dining at restaurants, and shopping at retail stores. These changes in movement patterns can greatly affect the emergency call volumes of fire departments. If the social distancing measures are effective, one can expect that emergency incidents related to COVID-19 symptoms will eventually decrease over time. However, the reduced mobility can also affect the frequency of other incident types such as traffic accidents and injuries. In this blog post, we will examine the impact of social distancing on emergency call volume for two departments that utilize the National Fire Operations Reporting System (NFORS) data system. The two departments include Miami-Dade Fire Rescue Department in Florida (*which does not include the City of Miami) and the Columbus Division of Fire in Ohio.</p><h3>How can we quantify how much social distancing a community is doing?</h3><p>In order to help researchers and public health officials understand the impact of social distancing in their communities, Google has released <a href="https://www.google.com/covid19/mobility/">Community Mobility Reports</a>. These reports provide aggregated and anonymized data collected from mobile phones whose users have opted into the Location History setting. At the county level, the dataset provides the percent change in visits to various types of places relative to a baseline. The baseline is calculated from the daily number of visits observed before the U.S. implemented widespread measures in response to the pandemic. In this blog post, we explore four location types from the mobility dataset:</p><p><strong>Grocery &amp; pharmacy- </strong>Mobility trends for places like grocery markets, food warehouses, farmers markets, specialty food shops, drug stores, and pharmacies.</p><p><strong>Retail &amp; recreation- </strong>Mobility trends for places like restaurants, cafes, shopping centers, theme parks, museums, libraries, and movie theaters.</p><p><strong>Residential- </strong>Mobility trends for places of residence.</p><p><strong>Workplaces- </strong>Mobility trends for places of work.</p><h3>How is social distancing affecting total call volumes?</h3><p>First we examine the impact of mobility changes on total call volumes. The total call volume is based on the daily count of all incidents reported in NFORS. For simplicity, we show the call volume as a percentage change relative to baseline. The baseline was computed as the average number of calls per day from January 1st to February 6th. To make it easier to visualize the trends, we apply a moving average filter with a window of seven days to both the daily call volume data and the mobility data. This filter removes the noise caused by day-to-day fluctuations in these quantities. For example, any point shown on a day is actually the average value of that day, plus the three days before it, and the three days after it. This approach also removes the impact of the day-of-the-week because every point is an average of all seven days of the week. The resulting plot is shown below for Miami-Dade Fire Rescue Department, along with the changes in mobility for Miami-Dade County.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/648/0*Stzmw1gML6rxSmT6" /></figure><p>The same plot is generated for the Columbus Division of Fire using mobility data from Franklin County, shown below:</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/648/0*SC1VQdj-1uTbH4FB" /></figure><p>Both of these plots reveal similar trends. Here are some key takeaways:</p><ul><li>There has been a noticeable downward trend in total call volume resulting in 20% fewer calls than before social distancing measures were implemented.</li><li>Both communities began reducing visits to retail/recreation locations and workplaces around March 10th, which is about two weeks before shelter-in-place orders were implemented.</li><li>There was a spike in visits to grocery/pharmacy locations from March 8th to March 17th. After that time, these visits have been about 20–30% lower than the baseline.</li></ul><h3>Which types of call volume are most affected by social distancing?</h3><p>After seeing that total call volumes have declined, we were curious about which types of calls were driving these trends. We examined many different incident types for each department, and found that the most notable changes were in calls potentially related to COVID-19 symptoms, traffic accidents, and injuries. Because incident types are reported differently for each department, we used two slightly different approaches to identify these particular incidents.</p><p>Here are the filters we used to identify the incident categories for Miami-Dade Fire Rescue Department:</p><p><strong>Traffic calls- </strong>Any incident with the string “traff” in its type on the Computer Aided Dispatch (CAD) system. This string captures incident types that have either “traffic” or the abbreviation “traff.”</p><p><strong>Injury calls- </strong>Any incident with the string “injur” in its type. This string captures incidents that have either “injury” or “injuries” in their types. This string appears to exclude traffic related incidents.</p><p><strong>Fire calls- </strong>Any incident with the string “fire” in its type.</p><p><strong>Calls with COVID-19 keywords- </strong>Although Miami has a “FEVER WITH TRAVEL” incident type that was introduced in light of the recent pandemic, we were interested in all calls that could be related to COVID-19 symptoms, even if the patient did not recently travel. As a result, we identified incidents that have at least one of the following strings in the comments: “corona virus”, “coronavirus”, “cough”, “fever”, “covid.”</p><p>The resulting plot is shown below. Because there is not a meaningful baseline for “Calls with COVID-19”, we instead show the daily number of calls rather than a percent difference relative to a baseline. In order to show the mobility trends, we show the percent change relative to the baseline for visits to workplaces with the left axis. All other curves are plotted with the right axis.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/648/0*kvBYJpOivh8yu3Lk" /><figcaption>The mobility trend for workplaces (green curve) is plotted as a percent difference relative to the baseline using the left axis. All other curves represent the number of calls of that type per day and are plotted with the right axis.</figcaption></figure><p>For the Columbus Division of Fire, we used slightly different filters due to differences in reporting:</p><p><strong>Traffic calls- </strong>Any incident with the type “VEHICLE ACCIDENT.”</p><p><strong>Injury calls- </strong>Any incident with the string “injur” in its type. This string captures incidents that have either “injury” or “injuries” in their types.</p><p><strong>Fire calls- </strong>Any incident with the string “fire” in its type.</p><p><strong>CE calls- </strong>Any incident with the type “CONTAGIOUS EMERGENCY.” This is a new incident type that was introduced in the wake of the COVID-19 pandemic. This is why there are no calls of this type before March 12th. For more information, we recommend reading this <a href="https://medium.com/ipsdi/case-study-coronavirus-covid-19-dashboard-in-nfors-statengine-613928d0c267">blog post</a>.</p><p>The resulting plot is shown below. The apparent uptick in fire calls that occurs between March 17th and March 24th is because of the unusually large call volume on March 20th due to storms.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/648/0*e-quTeAv1mCOLfVZ" /><figcaption>The mobility trend for workplaces (green curve) is plotted as a percent difference relative to the baseline using the left axis. All other curves represent the number of calls of that type per day and are plotted with the right axis.</figcaption></figure><p>Looking the trends in the incident types for these two fire departments, here are some takeaways:</p><ul><li>Visits to workplaces have been near 50% of the baseline for both communities since the end of March. The most recent data for April shows a downward trend in the number of incidents with COVID-19 related keywords (for Miami-Dade) and incidents with the “CONTAGIOUS EMERGENCY” type (for Columbus). This trend may be evidence that the social distancing measures are working.</li><li>The reduced mobility appears to be linked to fewer injuries and traffic incidents. This trend is likely because people are driving less and avoiding other risky activities. These trends are more significant for Miami-Dade than for Columbus (Franklin County), however.</li><li>Fire call volume does not appear to have changed significantly for either department.</li></ul><h3>Conclusions</h3><p>In this blog post, we use data from NFORS and Google’s Community Mobility Report to examine the impact of social distancing on emergency call volumes. Although there is a general downward trend in total call volumes, an investigation of the individual incident types reveals a deeper story. We see that calls potentially related to COVID-19 increase sharply in mid-March, but the most recent data show that the frequency of these incidents is beginning to decrease. The reduced mobility also appears to be linked to a reduction in the frequency of incidents related to traffic accidents and injuries. The call volume for fire related incidents appears relatively unchanged. Going forward, these data can help departments monitor the impact of easing restrictions on their call volumes.</p><p>If your organization participates in NFORS and you would like to see a similar analysis for your department contact us today at <a href="/service/mailto:hello@i-psdi.org">hello@i-psdi.org</a> . If your department does not yet participate in NFORS, schedule a demo to see the power of analytics <a href="https://www.eventbrite.com/e/nfors-demo-registration-45782334194">https://www.eventbrite.com/e/nfors-demo-registration-45782334194</a> .</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=a17f248f224c" width="1" height="1" alt=""><hr><p><a href="https://medium.com/ipsdi/exploring-the-impact-of-social-distancing-on-emergency-call-volume-using-googles-mobility-dataset-a17f248f224c">Exploring the impact of social distancing on emergency call volume using Google’s mobility dataset</a> was originally published in <a href="https://medium.com/ipsdi">International Public Safety Data Institute</a> on Medium, where people are continuing the conversation by highlighting and responding to this story.</p>]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[NFORS: New Covid-19 base dashboard available]]></title>
            <link>https://medium.com/ipsdi/nfors-new-covid-19-base-dashboard-available-757ddb37ea8c?source=rss----ccb76b4bdcdd---4</link>
            <guid isPermaLink="false">https://medium.com/p/757ddb37ea8c</guid>
            <category><![CDATA[dashboard]]></category>
            <category><![CDATA[nfors]]></category>
            <category><![CDATA[covid19]]></category>
            <category><![CDATA[statengine]]></category>
            <category><![CDATA[data-visualization]]></category>
            <dc:creator><![CDATA[Austin Anderson]]></dc:creator>
            <pubDate>Mon, 27 Apr 2020 15:41:35 GMT</pubDate>
            <atom:updated>2020-04-27T15:41:35.465Z</atom:updated>
            <content:encoded><![CDATA[<figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/0*C1LQ2pPUzNPEJUsw" /><figcaption>Photo by <a href="https://unsplash.com/@dead____artist?utm_source=medium&amp;utm_medium=referral">Capturing the human heart.</a> on <a href="https://unsplash.com/?utm_source=medium&amp;utm_medium=referral">Unsplash</a></figcaption></figure><p>We have developed a new base dashboard that provides several useful visualizations to serve as a start to our users’ examination of their COVID-19 call data. The dashboard is available when creating a new workspace. As we have discussed <a href="https://medium.com/statengine/case-study-coronavirus-covid-19-dashboard-in-nfors-statengine-f9093a73b42a">in an earlier post</a>, <strong>deciding how to identify your COVID-19 calls is key</strong>. We cannot make this determination for our users as each user has a different situation.</p><h4>Features</h4><p>The new coronavirus dashboard features some familiar visualizations and some less common ones. The following discussion is under the assumption that you have determined the filters needed to properly identify your COVID-19 calls:</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*2PyoOnQyAcyTNWM424KigA.png" /><figcaption>Covid-19 base dashboard, available by creating a new workspace and choosing to include it.</figcaption></figure><ul><li>Incidents — Supplies the total incidents over the time period in question</li><li>Incident Map Gridded — A geospatial heat map of incident counts</li><li>Yesterday vs. 3 days ago — This visualization provides the percentage change in the total number of COVID-19 calls between the total number of calls on the most recent date in the date-time picker, and the total number of calls from 3 days before. For example, if on 04–04–2020 you had responded to 30 COVID-19 calls, and on 04–01–2020 you had responded to 60 calls, then the value displayed would be calculated as follows:<br>(((30–60)/30)-1)*100 = -100%<br>In other words, your calls on 04–04–2020 are 100% of 04–04–2020’s total call volume <em>less</em> than your calls on 04–01–2020.</li><li>Yesterday vs. 3 days ago line — This is a line graph representation of the above metric, displaying historic trends.</li><li>Incidents by First Due — The number of COVID-19 calls coming in from the various first due areas of the department. This may help identify areas with unusual concentrations of COVID-19 calls.</li><li>Average Calls per Day — The number of incoming COVID-19 calls per day.</li></ul><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*Tp_ZCtxC_0-TRkM1iKczIA.png" /><figcaption>Additional view of COVID-19 base dashboard, displaying additional metrics.</figcaption></figure><ul><li>Key metrics — A summary of key department performance metrics, the 90th percentiles of First Turnout, First Travel, and Total Event (event open to event close) times</li><li>Incidents by Units — The number of COVID-19 calls responded to by each unit. This may provide insights into which units are most exposed to COVID-19.</li><li>Incident Comments — A table displaying the individual incident information, along with that incident’s dispatcher comments, and then the resources assigned to that incident, including units and personnel. This will help with detailed examinations of calls, and may serve as a place to help you tweak your filters for COVID-19 calls.</li></ul><h4>Conclusion</h4><p>This new dashboard is constructed to provide some initial visualizations that should provide a good foundation to our users who are addressing COVID-19 concerns. We want to emphasize again that <strong>good data collection</strong> <strong>is key</strong>. If your department has not yet determined a set method of classifying COVID-19 calls, be it through a dispatch type or through dispatcher comments, now is the time to do so! We believe such information will be immediately actionable, but even if not, it could serve as an important source of information for your city and other agencies examining the spread and impact of this virus.</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=757ddb37ea8c" width="1" height="1" alt=""><hr><p><a href="https://medium.com/ipsdi/nfors-new-covid-19-base-dashboard-available-757ddb37ea8c">NFORS: New Covid-19 base dashboard available</a> was originally published in <a href="https://medium.com/ipsdi">International Public Safety Data Institute</a> on Medium, where people are continuing the conversation by highlighting and responding to this story.</p>]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[Case Study: Coronavirus (COVID-19) Dashboard in NFORS/StatEngine]]></title>
            <link>https://medium.com/ipsdi/case-study-coronavirus-covid-19-dashboard-in-nfors-statengine-613928d0c267?source=rss----ccb76b4bdcdd---4</link>
            <guid isPermaLink="false">https://medium.com/p/613928d0c267</guid>
            <category><![CDATA[nfors]]></category>
            <category><![CDATA[public-safety]]></category>
            <category><![CDATA[covid19]]></category>
            <dc:creator><![CDATA[Austin Anderson]]></dc:creator>
            <pubDate>Thu, 19 Mar 2020 16:17:12 GMT</pubDate>
            <atom:updated>2020-04-10T20:52:03.961Z</atom:updated>
            <content:encoded><![CDATA[<figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/0*WwFn32C5ScOWkzwB" /><figcaption>Photo by <a href="https://unsplash.com/@cdc?utm_source=medium&amp;utm_medium=referral">CDC</a> on <a href="https://unsplash.com/?utm_source=medium&amp;utm_medium=referral">Unsplash</a></figcaption></figure><p><strong>Update</strong> (04/10/2020): We have added <a href="https://medium.com/statengine/nfors-new-covid-19-base-dashboard-available-9f8910de5d4">a new base dashboard</a> in NFORS to give our users a good foundation to start examining their COVID-19 incidents!</p><p>With first responder agencies everywhere now receiving COVID-19 calls and with the need to track them closely both in cooperation with other agencies and for the purposes of monitoring personnel exposure to the virus, we wanted to present some recommendations on setting up a COVID-19 dashboard. This request came from one of our NFORS users asking for assistance to build and revise a dashboard for tracking their coronavirus responses.</p><h4>Good data quality is the key</h4><p>The most important key to creating a good NFORS dashboard for COVID-19 is to ensure that your department has a good way to designate COVID-19 call types in your CAD dispatch. When we first started helping our user department with building their dashboard, they were sending COVID-19 calls out on typical EMS dispatch types, such as DIFFICULTY BREATHING or ILL PERSON, with additional notes in the dispatcher comments to designate these calls using either coronavirus, COVID-19, or PPE advised. After 2 or so days of this initial “ad-hoc” phase, they <strong>added a new call type to their dispatch for COVID-19 calls, CONTAGIOUS EMERGENCY</strong>. This change ensured that they could easily filter for and track their coronavirus incidents. Thus, <strong>our recommendation to NFORS users is to contact your CAD administrators ASAP and implement a call type specifically for your COVID-19 calls</strong>.</p><p>During the initial phase before they implemented the CONTAGIOUS EMERGENCY call type, we were able to help them get a handle on their incidents by filtering their incidents by text in their comments using the following procedure:</p><ol><li>In a dashboard, add a filter on description.comments containing the keywords being used by your dispatchers to designate COVID-19 runs, such as covid, coronavirus, or even just “ppe advised”, and save it.</li><li>Select “Edit filter,” and in the “Add filter” dialog that pops up, select “Edit Query DSL”</li><li>Delete the line with the word “phrase” on it, as well as the comma on the line above it.</li></ol><figure><img alt="" src="https://cdn-images-1.medium.com/max/652/1*cs5nYmDSvFpdorqciF4O9Q.gif" /><figcaption>Creating the comments filter</figcaption></figure><figure><img alt="" src="https://cdn-images-1.medium.com/max/652/1*1CtpkRZkUbD4E4I4XaUpJw.gif" /><figcaption>Proper editing of the comments filter. This small edit causes the comments test search filter to be case-insensitive.</figcaption></figure><h4>Adding raw COVID-19 calls data to a dashboard</h4><p>It proved important to our users to examine the raw incident data from their dashboard, so that they would have a full view of the incidents they were examining based on other filters in their dashboard.</p><p>Normally, the “discover” tab in NFORS is used to examine raw incident data. However, with the “saved searches” functionality, we can develop a filtered view of the raw incidents and then add that filtered view to our dashboard! Here’s how:</p><ol><li>In the “discover tab” add the filters necessary to capture your department’s COVID calls, perhaps even using the trick outlined above.</li><li>At the dashboard, select “Edit” followed by “Add,” then choose the “Saved Searches” tab and choose your COVID calls saved search.</li></ol><figure><img alt="" src="https://cdn-images-1.medium.com/max/832/1*352LXBzXoVV5Lx6-zUzZgw.gif" /><figcaption>Saving the COVID-19 search from the discover tab</figcaption></figure><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*jhauvfG0bmeIpzCdHphXwQ.gif" /><figcaption>Adding a saved search for COVID-19 calls</figcaption></figure><h4>Useful Visualizations for your COVID-19 dashboard</h4><p>Now that you have the way to filter for your incidents as well as see the raw incident data in your dashboard, we would like to list several default visualizations that our user found useful in building their COVID-19 dashboard. <strong>Keep in mind that these visualizations should all be filtered for COVID-19 calls according to the strategies outlined previously</strong>.</p><ul><li>Incidents — This will give you the total number of COVID-19 incidents to which your department has responded over the time period of interest.</li></ul><figure><img alt="" src="https://cdn-images-1.medium.com/max/302/1*c6lEc5bL5-G3YWDGIAmrcg.png" /><figcaption>Note: Data above is synthetic.</figcaption></figure><ul><li>Incidents by Time — Provides the trend of COVID-19 incidents to which your units are responding.</li></ul><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*1ldbJpVRQ0-dxu2ebORIUw.png" /></figure><ul><li>Incident Map — A map of all your COVID calls. Remember that the tools on the top left side of the map allow you to filter regions of interest.</li></ul><figure><img alt="" src="https://cdn-images-1.medium.com/max/491/1*VmZ7rsjDLcgFBNAdDMynXQ.png" /><figcaption>Note: This is synthetic data for Richmond, not actual COVID-19 calls.</figcaption></figure><ul><li>90th percentile EMS Turnout/Response/Total event duration — For tracking your unit performance on COVID-19 calls, as well as having an idea of how long units are spending committed to the incident.</li></ul><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*CkLokZy1eDKe8UK0exmACg.png" /></figure><ul><li>Incidents by Shift — Used by the department in conjunction with the time picker to examine the total number of calls taken by a given shift in a given period of time, to aid in staff planning and monitoring personnel exposure to the coronavirus.</li></ul><figure><img alt="" src="https://cdn-images-1.medium.com/max/399/1*s5h-jKc2xmAjT_OE_kGFvA.png" /></figure><ul><li>Incidents by Unit and Category — This will detail which units are responding to the largest number of COVID-19 calls, an indication of exposure.</li></ul><figure><img alt="" src="https://cdn-images-1.medium.com/max/838/1*z5PmwCpzvOziSQaf3YMXVA.png" /></figure><ul><li>Common Addresses with Census Tract Characteristics — Supplies at-a-glance the addresses where your units are responding to COVID-19 cases, with some demographic characteristics based on Census data.</li></ul><figure><img alt="" src="https://cdn-images-1.medium.com/max/908/1*ZtvnkoNlDJVli4OtyDWYvQ.png" /></figure><p>And many more according to whatever you may need!</p><h4>Summary</h4><p>The most important need for making a useful COVID-19 dashboard is being able to distinguish which incidents are COVID-19 calls and filtering for them. If possible, <strong>work with your CAD administrator to establish a dispatch type exclusively for COVID-19 calls</strong>. If such an arrangement cannot be made, then try to at least get mentions of COVID-19 or coronavirus into the dispatcher comments in a consistent way so that you can find those incidents later. Once you have a handle on identifying which incidents are COVID-19 calls and filtering for them, then even existing dashboards such as our default “Incident Dashboard” will supply a wealth of important information related to your department’s COVID-19 calls.</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=613928d0c267" width="1" height="1" alt=""><hr><p><a href="https://medium.com/ipsdi/case-study-coronavirus-covid-19-dashboard-in-nfors-statengine-613928d0c267">Case Study: Coronavirus (COVID-19) Dashboard in NFORS/StatEngine</a> was originally published in <a href="https://medium.com/ipsdi">International Public Safety Data Institute</a> on Medium, where people are continuing the conversation by highlighting and responding to this story.</p>]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[How NFORS data can help departments decide where to house new response units]]></title>
            <link>https://medium.com/ipsdi/how-nfors-data-can-help-departments-decide-where-to-house-new-response-units-ca903f802aca?source=rss----ccb76b4bdcdd---4</link>
            <guid isPermaLink="false">https://medium.com/p/ca903f802aca</guid>
            <category><![CDATA[public-safety]]></category>
            <category><![CDATA[nfors]]></category>
            <category><![CDATA[data-science]]></category>
            <dc:creator><![CDATA[Tyler Buffington, PhD]]></dc:creator>
            <pubDate>Wed, 18 Mar 2020 13:01:29 GMT</pubDate>
            <atom:updated>2020-03-18T13:01:29.079Z</atom:updated>
            <content:encoded><![CDATA[<p><em>Written by </em><a href="https://medium.com/@tyler.c.buffington"><em>Tyler Buffington</em></a><em>, </em><a href="https://medium.com/@garnertb"><em>Tyler Garner</em></a><em>, &amp; </em><a href="https://medium.com/@joe.chop"><em>Joe Chop</em></a></p><p>In this blog post, we will explore how the operational data collected in NFORS can help inform decisions involving the placement of new units in a fire department. As an example, we will analyze data from Delray Beach Fire Rescue to provide quantitative insight into the question:</p><blockquote>“What would be the impact of adding an additional ALS unit to a station?”</blockquote><h3>First due vs. non-first due responses</h3><p>Fire departments commonly divide up their jurisdiction into “first due areas.” These areas identify the optimal fire station response to any given location. When a unit is required at an emergency incident location within a station’s first due area, it is generally deployed from the station. If the required unit type is not available at the first due station, then another station outside the first due area must send a unit instead. This response from outside a station first due area generally results in an increased response time to the incident location. The graphic below shows the division of first due areas for Delray Beach Fire Rescue.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/665/0*9CrjeumXIUyr-SiH" /></figure><p>As EMS calls usually account for the majority of calls for service, the analysis will focus on this type of calls. In order to do this, we assess every incident when an ALS unit was dispatched in the available data. Based on the address of the incident, we can determine which station should be first due. If the unit belongs to the first due station, then we call this a dispatch from the first due station. Otherwise, it was a dispatch from a non-first due station. A cumulative distribution function (cdf) plot is useful for comparing the response time distributions of the two types of ALS dispatches, which is shown below. The faster the line “climbs” to the 100th percentile, indicates the higher frequency of responses from that type station (first due vs. non first due).</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/432/0*Wkq8OjAuoKZoB1XE" /></figure><p>The above plot shows the unit response time plotted on the x-axis and the corresponding percentile (or quantile) on the y-axis. The percentile refers to the percentage of unit response times that fall below the response time on the x-axis. For example, we can see from the plot that roughly 30% of responses from first due stations are faster than 4 minutes, and roughly 30% of responses from non-first due stations are faster than 6 minutes. We can also see that 90% of responses from first due stations fall below 8 minutes, but only 60% of responses from non-first due stations fall below 8 minutes. The difference in medians (the 50th percentiles) is also about 2 minutes with first due responses being faster. This visualization is useful because it shows the comparison of the entire response time distributions as opposed to a comparison of a single quantity such as the mean, which emphasizes the statistical nature of response time. The two curves can be thought of as two different lotteries. It is possible to “draw” a short response time from the red distribution and it is possible to “draw” a long response time from the blue curve, but on average, the blue curve is a better lottery.</p><h3>Breaking it down by first due area</h3><p>It may also be useful to examine the consequences of non-first due dispatches for each of the separate first due areas. This is shown in the plots below:</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/864/0*OyQz3AxpExK_hd6r" /></figure><p>As expected, the response time distribution is better for all units deployed from station to incidents in their first due areas. However, the above plots show that the consequences associated with non-first due dispatches are more severe in some regions than others. This essentially comes down to how well other stations are able to “fill in” for a given first due station. Most of the station locations in the more populated regions of the city are close to multiple stations. In other words, their first due areas are smaller geographically speaking. If an incident occurs in these regions and the first due station cannot respond, it is likely that another station can respond in a relatively timely manner. Conversely, Station 116 has a larger geographic first due area to cover, meaning that if Station 116 is unable to respond, the next closest unit must travel a much longer distance. At first glance, this may seem to indicate that Station 116 would be the best place to add a new unit. However, up until this point, our analysis has not considered two other important factors of an emergency response system — which are first due reliability and simultaneous call volume.</p><h3>First due reliability and factors that influence it</h3><p>As previously stated, a major consideration for deciding which station should house a new unit is the station’s reliability. Here we define the station’s reliability as <em>the fraction of units dispatched to incidents in a station’s first due area that are sent from the first due station. </em>Imagine that an ALS unit is required at an incident location in the first due area of Station X. What is the probability that Station X sends the ALS unit as opposed to some other station? That is the idea behind the first due reliability. This quantity is important because a station with a high first due reliability can offset the consequences associated with non-first due dispatches.</p><p>A major factor that influences the first due reliability of a station is the number of units it houses. Obviously a unit cannot respond to more than one call at a time. For Delray Beach Fire Rescue, we observe that 65% of non-first due dispatches for ALS units occur while all of the respective units from the first due station are already responding to other incidents. Additionally, the tendency for multiple incidents to occur simultaneously in the first due area also affects the first due reliability of a station. <strong>Using this information, we can model the impact of adding new units as an improvement to the first due reliability of the stations that house them</strong>.</p><h3>Putting it all together and understanding tradeoffs</h3><p>The last consideration that we explore is the overall call volume for each of the stations’ first due areas. Stations with large call volumes in their first due areas greatly influence the overall response time distribution of the department. The tree diagram below shows all six stations of Delray Beach Fire Rescue and the metrics we’ve discussed that influence the potential impact of adding a new ALS unit including call volume, first due reliability, and response time.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/884/0*7SQd5CqRXnJ6uanU" /></figure><p>Moving from left to right in the tree diagram, the first column contains the percentage of call volume for each of the first due areas. For example, 27% of all ALS units dispatched are responding to incidents in the first due area of Station 111. Of those units, 66% of them belong to Station 111. The 90th percentile response time of ALS units belonging to Station 111 responding to incidents in the first due area of Station 111 is 7 minutes and 32 seconds. Note that the 90th percentile is displayed here for simplicity; in reality, the response time modeled as a random draw from a distribution.</p><p>The tree diagram is useful for showing the tradeoffs between stations that could affect the impact of adding additional units. For example, the first due area of Station 116 exhibits the largest average difference in response times between first due and non-first due dispatches, but ALS units are only sent there 4% of the time. On the other hand, Station 111’s first due area has a much smaller average response time difference, but it has 27% of the call volume and only 66% of the units sent to this region belong to Station 111.</p><h3>Modeling tradeoffs</h3><p>As previously mentioned, the historical data can be used to generate rough estimates of the impact of a new ALS unit at each station. At a high level, these estimates are calculated by identifying cases in which the first due station could not respond because all of its units were busy, but it is likely that the station could have responded if it had one more unit. For these cases, we assign hypothetical response times based on the “from first due” response time distribution for that station. The table below shows an example of the results produced from this model.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/292/0*XKDNbKSmhXn7gReq" /></figure><p>As shown above, the model predicts that the best place for a new ALS unit is Station 111, which is the busiest station. The model expects that adding a new unit to this station would save <strong>1,653 minutes</strong> of response time per year. For this department, the expected improvement follows a similar trend for the stations’ call volume. This measure may be different for other departments depending on their station’ call volumes, first due reliability, and delays due to non-first due dispatches.</p><h3>Conclusions</h3><p>In this blog post, we explore how NFORS data can provide insight to a department facing a decision of whether to purchase a new unit and where a new unit should be placed. We examined three metrics that can be calculated from NFORS data that can influence the potential impact of adding a new unit to a station:</p><ul><li>How often units are required in a station’s first due area (call volume)</li><li>How often the station is able to respond to calls in its first due area</li><li>The difference in response times between cases when the first-due station is able to respond and when other stations must send units instead</li></ul><p>Lastly, we briefly introduced a model that is able to identify cases in which the first due station could not respond because all of its units were responding to other calls when an additional unit was required. For some of these cases, the first due station would have been able to respond if it had an additional unit. For these cases, we can replace the actual response times with response times from cases in which the first due station could respond. This allows for the calculation of an expected improvement in response times associated with adding a new unit to each station in a department. It is important to note that this analysis may vary if your department uses proximity based dispatching (closest unit).</p><p>If your organization participates in NFORS and you would like to see a similar analysis for your department contact us today at <a href="/service/mailto:hello@i-psdi.org">hello@i-psdi.org</a></p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=ca903f802aca" width="1" height="1" alt=""><hr><p><a href="https://medium.com/ipsdi/how-nfors-data-can-help-departments-decide-where-to-house-new-response-units-ca903f802aca">How NFORS data can help departments decide where to house new response units</a> was originally published in <a href="https://medium.com/ipsdi">International Public Safety Data Institute</a> on Medium, where people are continuing the conversation by highlighting and responding to this story.</p>]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[A Guide to Analyzing Fire Department Response Coverage with NFORS Data.]]></title>
            <link>https://medium.com/ipsdi/a-guide-to-analyzing-fire-department-response-coverage-with-nfors-data-a1e9ca5879f2?source=rss----ccb76b4bdcdd---4</link>
            <guid isPermaLink="false">https://medium.com/p/a1e9ca5879f2</guid>
            <category><![CDATA[data-visualization]]></category>
            <category><![CDATA[fire-department]]></category>
            <category><![CDATA[nfors]]></category>
            <category><![CDATA[public-safety]]></category>
            <category><![CDATA[kernel-density-estimation]]></category>
            <dc:creator><![CDATA[Tyler Buffington, PhD]]></dc:creator>
            <pubDate>Mon, 22 Jul 2019 19:49:52 GMT</pubDate>
            <atom:updated>2019-07-22T19:49:52.372Z</atom:updated>
            <content:encoded><![CDATA[<h3>Using NFORS data to understand coverage.</h3><p><em>Written by </em><a href="https://medium.com/@tyler.c.buffington"><em>Tyler Buffington</em></a><em> &amp; </em><a href="https://medium.com/@garnertb"><em>Tyler Garner</em></a><em> — July 22, 2019</em></p><p>Fire Chiefs and other users often ask how the real-time operational data collected in NFORS can help inform decisions regarding the geographic placement of new resources. For example:</p><blockquote>Can the NFORS data determine the ideal location for an additional fire station or apparatus placement?</blockquote><p>These decisions are quite complicated and depend on a multitude of factors including land availability, incident prevalence, projected population growth and demographic trends, and road networks. Additionally, the impact of these decisions can be considerable.</p><p>In his book, “<em>The Fires: How a Computer Formula, Big Ideas, and the Best of Intentions Burned Down New York City — and Determined the Future of Cities</em>,” Joe Flood discusses perhaps the most well known example of misguided resource allocation decisions leading to severe consequences. He describes how the results from computer models led to the closure of key fire stations in New York, which contributed to an epidemic of severe fires in the 1970s. This example highlights the need for decision-makers to rely on their experience and knowledge of their own community in addition to data-driven models to inform resource allocation decisions. Putting the data models in the context of a particular community leads to clear information and enlightenment for decision making.</p><p>With the understanding of data in context, the large volume of operational data collected from NFORS can help departments better understand the coverage profile of their community and aid in this type of decision making. In this blog post, we describe several informative visualizations that can help decision-makers better understand their community’s specific needs, as well as how the department is currently serving those needs. Let’s start by reviewing historic travel times to see if certain locations are underperforming.</p><p>At the most basic level, we can create a scatter-map of historic incidents and the corresponding first unit travel time. An example is shown below, with the 0 on the X axis corresponding to the longitude of the furthest west incident and the 0 on the Y axis indicating the latitude of the furthest south incident. Each point represents an incident and its color is determined based on the travel time of the first unit. <em>n= 5,410</em> incidents shown.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/705/0*MDUgdZm1lUSkw4z9" /></figure><p>Although this visualization conveys a lot of information, it’s difficult to interpret, especially since it lacks geographic features which would provide an important contextual reference. It is also difficult to identify trends in regions with a large number of historic incidents since multiple incidents may be stacked in a single location. Perhaps a more informative visualization is an incident density heat map.</p><p>To generate this visualization, we employ a technique called Kernel Density Estimation. The technical details of using this methodology are outside the scope of this article, however the resulting visualization gives an indication of how many emergency calls are distributed throughout this department’s service area. Also shown is the department’s jurisdictional boundary and station locations.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/747/0*aVNKXcnQacPp4H8a" /></figure><p>This map is useful for visualizing the geographic demand profile. The areas that are shaded red indicate the area where a large number of incidents have occurred. Compare this visualization to the initial scatter map and although it is possible to see “clusters” of incidents, it is difficult to interpret the actual call density since many incidents were in close proximity.</p><p>However, it is important to note that the incident density visualization does not provide any information about travel times and therefore, the user does not have any insight into a critical performance component of these locations.</p><p>To help NFORS users understand locations that have historic travel time challenges, we can use a kernel smoother to calculate travel times to each location in the service area. In this visualization, each colorized location represents a weighted average. Incidents closer to one another on the map are assigned larger weights because they are assumed to be more “similar.” The results are somewhat intuitive, with incidents further away from a resource location typically having longer travel times.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/725/0*-5NomPnlXJT6Q-ab" /></figure><p>It is important to note that we only assign colors in this plot to regions with more than 20 historical incidents within a <strong>½ </strong>mile radius. We do this because there are not enough data to estimate average travel times in the regions with fewer incidents. Therefore, some regions within the department’s jurisdiction do not have any color assignment.</p><p>Now that we’ve analyzed high travel times and high call volume, how can we combine these visualizations to produce a single visualization that gives the user insights into the locations where additional resources may be needed?</p><p>To assess these areas, we can combine the previous visualizations to identify locations where <em>both</em> travel times are slow <em>and</em> call volumes are high as shown in the diagram below.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/0*7oerS7Wx34MVsEiJ" /></figure><p>To accomplish this, we multiply the quantiles of the two previous heat maps together to generate a third visualization. In this map, regions that are red indicate <em>both</em> large call volumes and long travel times. It is important to note that this approach assumes an equal weighting of call volume and travel time. Depending on the situation, a decision maker may wish to place more weight on regions of high call volume over regions of slow travel times or vice versa.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/743/0*-ifbMfu9Qenqya9i" /></figure><p>In addition to our current analytics, NFORS is currently developing new visualization tools like this to help departments better understand their operational data. For example, using NFORS department leaders could generate the visualizations outlined in this blog post, filtered by incident type (i.e. fire vs. EMS) or by time of day. We are also experimenting with new visualizations of incidents in which the first arriving unit was not from the first due station so that fire department leaders can better understand unit reliability and the impact of high volume in particular areas of their jurisdiction.</p><p>As an NFORS user, we welcome your ideas for new visualizations and tools that could help you better <strong><em>use</em></strong> <strong><em>your data</em></strong> <strong><em>to</em></strong> <strong><em>tell your story</em></strong>. If you would like to learn more, or have us create these visualizations for your department. Reach out to our professional services team at <a href="/service/mailto:hello@i-psdi.org">hello@i-psdi.org</a>.</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=a1e9ca5879f2" width="1" height="1" alt=""><hr><p><a href="https://medium.com/ipsdi/a-guide-to-analyzing-fire-department-response-coverage-with-nfors-data-a1e9ca5879f2">A Guide to Analyzing Fire Department Response Coverage with NFORS Data.</a> was originally published in <a href="https://medium.com/ipsdi">International Public Safety Data Institute</a> on Medium, where people are continuing the conversation by highlighting and responding to this story.</p>]]></content:encoded>
        </item>
    </channel>
</rss>