The Map View

The map view serves as the spatial foundation of the dashboard, plotting a point for every restaurant using bivariate symbology: colour encodes the restaurant’s most recent inspection grade, and circle size scales with its inspection score (a higher score means more, or more severe, violations, so worse-scoring restaurants read as larger circles at a glance). The default dashboard basemap is Esri’s Dark Gray Canvas, with an Esri World Imagery satellite toggle available in the top-right corner for real-world context. Hovering a restaurant point at zoom level 16 (map scale 1:18,056) or closer previews the restaurant’s name, grade, and score directly on the map; clicking a restaurant point loads its identifying and geographical details, full inspection history, and violations in the side panel, with its score-over-time chart shown below the map, and the restaurant listing panel gives an alternate way to browse locations where multiple restaurants share a point. The primary colours for Grade A, Grade B, and Grade C were derived from sampled NYC letter-grade signage. Grades N, P, and Z, each representing a restaurant still awaiting a finalized grade, are consolidated into a single Pending category; restaurants with no scored inspection on record appear as Uninspected, and restaurants closed by DOHMH appear as Closed. This map allows users to explore geographic patterns in restaurant performance and serves as the primary interface for interacting with and filtering the dataset.

A dedicated info button on the map panel opens a matching reference for its own legend, interactions, and data notes, part of the contextual documentation system covered below.

Map Controls

Custom controls, built to replace ArcGIS’s own map widgets, are arranged in the map’s four corners. The top-left cluster holds zoom in/out buttons and a compass that passively tracks the view’s rotation, resetting to north on click.

Below them, the Map Scale and Zoom Lvl readouts double as inputs: clicking either one turns it into a text field, letting you jump straight to an exact scale denominator or zoom level instead of scrolling to it.

A graphical scale bar in the bottom-right corner shows the current map scale as a ruler, converting the view’s resolution to true ground distance with a latitude correction and snapping to a round foot or mile value.

The top-right corner pairs a satellite/map basemap toggle with the Search Radius tool’s target icon. On phone and tablet layouts, a third “locate me” button stacks below them, using the device’s own location to centre the map and sort the restaurant list by distance; it’s hidden on desktop.

Activating Search Radius switches the cursor to point-placement mode; dropping a point re-scopes the restaurant list, stats panel, and grade breakdown chart to a 0.25, 0.5, or 1 mile circle around it instead of the full map view, with a distance field added to each list card.

Stats Panel (Current Map View)

The Stats Panel pairs a compact count bar with a donut chart: the bar gives the total restaurant count and its split across every grade category (A, B, C, Pending, Uninspected, and Closed), while the donut shows that same split as percentage shares. Both update dynamically with the current map extent, re-scope to the Search Radius circle when one is placed, and narrow to match any active grade filter, so the chart, list, and map always agree on what’s on screen.

Dashboard Documentation

Rather than one static instructions panel, the rebuilt dashboard documents itself contextually. A persistent Dashboard Information panel shows dataset freshness at a glance (last updated date, total restaurant and inspection counts, and attribution); its info button opens a full reference guide covering an overview of every feature, step-by-step how-to-use instructions, grade and status definitions, data-quality notes, and links to NYC’s own health-department resources. Every other panel, including the map, restaurant details, and stats panel, carries its own matching info button, opening guidance specific to that panel instead of one generic instructions sheet.

The Map View’s own info popup is one example of that per-panel pattern: the same overview, how-to-use, and data-notes structure, scoped to just that panel’s controls and legend.

Responsive Layout

Rather than one layout compressed to fit every screen, the dashboard switches between three distinct arrangements depending on viewport size, each rebuilt around what actually fits at that size.

Above 1750px, a three-pane layout puts the map, restaurant list, and stats panel and charts side by side, all visible at once.

Between 750px and 1750px, those panes stack into a scrolling two-column layout topped by a slim app bar, a compact wordmark plus Search, Filters, and Info buttons that each open as a drawer instead of taking up permanent space; the Grade Breakdown donut folds into a chevron drawer on the stats bar, and the Restaurant Performance chart runs full-width along the bottom.

Below 750px, a dedicated phone layout takes over, designed around a portrait phone: a slim top bar, a full-screen map, and a bottom sheet that steps between three heights, a peek, half, and full view, so the map stays the primary surface and the Restaurant Explorer only takes up as much space as it needs. Search, filters, and info each open as a drawer over the map here too. It isn’t built for landscape use, but a short landscape viewport (480px tall or less) keeps this layout anyway rather than switching to the two-column one, since there isn’t enough vertical room to stack panes there either.

Grade Breakdown — Map View

The Grade Breakdown is a donut chart titled “Grade Breakdown — Map View” by default, retitling itself “Grade Breakdown — Within [radius]” when the Search Radius tool is active. Each slice’s share of the total appears as a percentage beneath the chart, alongside the raw counts in the Stats Panel above it, so the same data reads as both an absolute count and a proportion.

Unlike the original ArcGIS Dashboards version, this chart is read-only: it no longer supports clicking a slice to filter the map, since grade and borough filtering now live in their own dedicated controls. Instead, it reflects whatever is already filtered or searched for, narrowing to just the selected categories when a grade filter is active, and to the matching restaurants when a borough filter or search term is applied. A “Filters applied” note beneath the chart names any active borough or search filter, since those narrow the tally without changing which slices are drawn.

Restaurant Explorer

Selecting a restaurant, on the map or in the list, opens three linked panels for digging into that single restaurant: a sorted list to find it, a details view summarizing it, and a full report for any one of its past inspections.

The Restaurant List shows every restaurant currently in scope (the map view, or the Search Radius circle when one’s placed), narrowed by any active Grade and Borough filters or search text. Each card carries the restaurant’s name, address, cuisine, most recent inspection date, and grade/score badges; selecting one opens its details and pans the map to its location. Results sort by up to two combinable keys (name, date, grade, score, cuisine, or distance) with a direction toggle, and paginate at 10 per page; once a Search Radius or “locate me” point is set, Distance becomes available as both a per-card readout and a sort key.

Restaurant Details

The Restaurant Details panel summarizes the selected restaurant itself: a Restaurant Information table (CAMIS, cuisine, current status), a Geographical Information table (borough, street, zipcode, phone, a location-verification badge, and a Google Maps link), and a full Inspection History list, one row per past inspection, each showing its date, grade, score, and critical/not-critical violation counts. Selecting a history row opens that inspection’s full report.

Inspection Reports

Inspection Reports show the complete record for one selected inspection, defaulting to the restaurant’s most recent. Each report lists the inspection type, the action taken, a “Closed by DOHMH” badge when that inspection resulted in a closure, and the full violation list with each code’s category and critical/not-critical status. Newer and older navigation buttons step between a restaurant’s inspections without returning to the Details tab.

Filter selections

Two filters, Grade and Borough, live together in the dashboard’s Filters control (an inline bar on wide desktop, a popover or drawer on compact and mobile layouts) and combine freely with each other and with the search field. Both are multi-select toggle buttons: turning on more than one option within a filter shows restaurants matching any of them, and each filter clears independently with its own Clear button.

The Grade filter toggles between six colour-coded categories: A, B, C, Pending (grades N, P, and Z), Uninspected (no scored inspection on record), and Closed (closed by DOHMH). Selecting one or more narrows the map, restaurant list, stats panel, and grade breakdown chart to just those categories.

The Borough filter narrows results to one or more of the five boroughs: Bronx, Brooklyn, Manhattan, Queens, and Staten Island. Unlike a Grade filter, an active Borough filter doesn’t reshape the Grade Breakdown chart’s own slices, since the chart always tallies every grade; instead it shows up in the “Filters applied” note beneath the chart, alongside any active search term.

Restaurant Performance Over Time

Restaurant Performance Over Time is a line chart plotting each of a restaurant’s scored inspections by date, drawn straight from its own inspection history rather than a separate scores dataset. Three shaded background bands mark the same grade thresholds used everywhere else in the dashboard, A (0–13), B (14–27), and C (28+), so a point’s position alone shows its grade at a glance. Each point is colour-coded to its own grade category rather than a flat colour, and inspections without a numerical score are excluded from the timeline.

The chart is interactive, not just illustrative: hovering a point, or tabbing to the chart and using the arrow keys, opens a tooltip with that inspection’s date, score, and grade; clicking a point, or pressing Enter or Space while it’s focused, switches to that inspection’s full report on the Inspection Reports tab. It’s linked to the Restaurant Details panel’s Inspection History list in both directions, hovering a history row previews its point here, and hovering a point highlights the matching row.

This chart is one of the strongest analytical components of the dashboard, providing a clear temporal perspective on restaurant performance that cannot be captured through static or aggregate views alone.

Patterns & Distribution

When visualizing the current dataset (30,566 restaurants), the same spatial trend holds up: Manhattan has the highest concentration, with 12,191 restaurants, about 40% of the citywide total, reinforcing its role as the city’s primary dining hub. Other boroughs exhibit lower relative counts, resulting in more dispersed restaurant distributions and more localized patterns.

While the restaurant counts vary significantly by borough, the trend does not appear to be correlated to population alone. For example, the estimated population of Manhattan in 2020 was 1,694,251—the second-lowest among the 5 boroughs (New York City Department of City Planning, 2024)—yet it accounted for 57.6% of NYC’s total tourism jobs (Office of the New York State Comptroller, 2021). This suggests that restaurant density is likely influenced by tourism and commercial activity, which increases demand for dining establishments beyond what the resident population alone would indicate.

The same pattern carries through the grade breakdown: boroughs with higher restaurant density, particularly Manhattan, also show higher counts across every inspection grade. This reflects overall restaurant volume rather than a performance difference, since higher density increases total inspection counts without indicating a decline in sanitary conditions.

Temporal Trends

A review of restaurant inspection histories using the Restaurant Performance chart reveals that performance is dynamic. While some establishments maintain relatively consistent scores across multiple inspections, others show noticeable fluctuations, with scores improving or worsening between inspections. Some restaurants trend gradually upward over time, meaning declining performance, since a higher score means more or more severe violations, as in the example below.

These variations suggest that compliance with food safety standards can change over time rather than remaining constant. Reviewing a restaurant’s full inspection history, rather than just its most recent result, provides context a single inspection can’t, and reinforces the value of tracking performance over time. Other restaurants show high variability instead of a steady trend, with scores swinging between inspections rather than moving consistently in one direction, as in the image below.

Interaction Insights

The dashboard’s interactive map directly shapes how the dataset reads: the stats panel, grade breakdown chart, and restaurant list all update dynamically to match the current map extent. As users pan and zoom across the city, the subset of data being visualized changes, revealing patterns that aren’t apparent at a city-wide view.

This relationship between map extent and the rest of the dashboard means interpretation varies with scale: broader views highlight citywide trends, while more focused views reveal finer spatial variation. The Grade and Borough filters, search field, and Search Radius tool narrow that same view further, letting users isolate specific subsets of the data without leaving the map.

Data & Dashboard Limitations

The following limitations were identified during the development and use of the dashboard:

• Multiple restaurants at the same or adjacent address often plot at the same map point, so a single click can only resolve to one restaurant even when several share it. The Stats Panel and Grade Breakdown chart aren’t affected by this, since both tally every restaurant currently in view from the underlying data rather than from what’s visually rendered as separate dots; the Restaurant List panel is the reliable way to browse and select between restaurants sharing a point.
• The dashboard has no backend or live database: restaurant data ships as one static GeoJSON file (all ~30,600 restaurants, about 32MB, loaded up front) plus over 27,000 small per-restaurant history files fetched on demand and capped at 50 entries in an in-memory cache. Static files keep hosting free and the app simple to reason about, since every filter, search, and sort runs entirely in the browser against data already downloaded, but a live database would scale further past this dataset’s size, at the cost of hosting fees and operational complexity a personal project doesn’t need.
• Data is only as fresh as the last build. DOHMH inspections are refetched dataset-wide at most once a day, driven by the scheduled geocoding backfill and any push to the main branch, not a live feed, so a newly published inspection can take up to a day to appear.
• LocationIQ’s free tier caps geocoding at about 4,900 lookups a day. A cold cache would take roughly a week to fully resolve; day to day this only matters for new or changed restaurants, since the daily backfill skips everything already cached.