28:null 29:["$","title",null,{"children":"Overture Places Quality Analysis | Wille Marcel | Observable"}] 2f:Tafb,{ let container = html`

Legend

`; // Give the container dimensions. yield container; // Create the \`map\` object with the mapboxgl.Map constructor, referencing // the container div let map = new maplibre.Map({ container, center: [-38.4587, -12.9924], zoom: 15.5, style: xyzStyle, scrollZoom: false }); map.addControl(new maplibre.NavigationControl(), "top-right"); map.on("load", () => { map.addSource("places", { type: "geojson", data: data }); map.addLayer({ id: "places-invalid", type: "circle", source: "places", paint: { "circle-radius": { base: 3, stops: [ [1, 4], [22, 180] ] }, "circle-color": "#fc8d59", "circle-opacity": 1 }, filter: ["==", ["get", "valid"], false] }); map.addLayer({ id: "places-valid", type: "circle", source: "places", paint: { "circle-radius": { base: 3, stops: [ [1, 4], [22, 180] ] }, "circle-color": "steelblue", "circle-opacity": 1 }, filter: ["==", ["get", "valid"], true] }); function popupContent(feature) { return ` ${feature.properties.name}
🏷️ ${feature.properties.category}
📈 ${Number(feature.properties.confidence).toPrecision(3)} `; } ["places-valid", "places-invalid"].forEach((layer) => { map.on("click", layer, (e) => { // Copy coordinates array. const coordinates = e.features[0].geometry.coordinates.slice(); const description = popupContent(e.features[0]); // Ensure that if the map is zoomed out such that multiple // copies of the feature are visible, the popup appears // over the copy being pointed to. while (Math.abs(e.lngLat.lng - coordinates[0]) > 180) { coordinates[0] += e.lngLat.lng > coordinates[0] ? 360 : -360; } new maplibre.Popup() .setLngLat(coordinates) .setHTML(description) .addTo(map); }); // Change the cursor to a pointer when the mouse is over the places layer. map.on("mouseenter", layer, () => { map.getCanvas().style.cursor = "pointer"; }); // Change it back to a pointer when it leaves. map.on("mouseleave", layer, () => { map.getCanvas().style.cursor = ""; }); }); // Be careful to clean up the map's resources using \`map.remove()\` whenever // this cell is re-evaluated. invalidation.then(() => map.remove()); }); }30:T71f,categoryGroups = [ { name: "Bars", categories: ["bar", "gay_bar", "eat_and_drink", "brewery", "beer_garden"] }, { name: "Cafes & Fast food", categories: [ "bakery", "eat_and_drink", "ice_cream_shop", "cafe", "cafetaria", "coffee_shop", "gelato", "desserts", "sandwich_shop" ] }, { name: "Medical facilities", categories: [ "hospital", "clinic", "doctor", "counseling_and_mental_health", "dentist", "physical_therapy", "health_and_medical", "naturopathic_holistic", "psychologist", "psychotherapist", "retirement_home", "laboratory_testing", "optometrist" ] }, { name: "Grocery Stores", categories: [ "supermarket", "health_food_store", "grocery_store", "organic_grocery_store", "cheese_shop", "candy_store" ] }, { name: "Veterinary and Petshops", categories: ["veterinarian", "pet_services", "pet_store"] }, { name: "Clothes", categories: ["women's_clothing_store", "clothing_store", "fashion"] }, { name: "Sport facilities", categories: [ "yoga_studio", "sports_and_recreation_venue", "sports_club_and_league", "martial_arts_club", "gym", "pilates_studio" ] }, { name: "Beauty facilities", categories: [ "beauty_and_spa", "barber", "beauty_salon", "spas", "hair_salon" ] }, { name: "Education", categories: [ "language_school", "specialty_school", "college_university", "day_care_preschool", "educational_research_institute", "school" ] }, { name: "Pharmacies", categories: ["pharmacy", "pharmaceutical_companies"] } ]31:[{"id":"6b488a7a83cb20c15578b9a4ebcb62d9d62f0cbc3ff28be2bf3cbe9d15e8e072ac61bec8ed8f79abd7adac571a73a15e040ba099345c0c6ec2d4710d80df32dd","href":"https://static.observableusercontent.com/files/6b488a7a83cb20c15578b9a4ebcb62d9d62f0cbc3ff28be2bf3cbe9d15e8e072ac61bec8ed8f79abd7adac571a73a15e040ba099345c0c6ec2d4710d80df32dd","name":"final@1.geojson","type":"application/geo+json","lastModified":1691081877676.321,"size":270663},{"id":"755e59c82b2dc06ddaa732c01ab343b52683a8f84392080432a7d216bd14829b3d2a762dee2ffe46430c66f298a9b40162d23e9ed5c7183098176a59b037413a","href":"https://static.observableusercontent.com/files/755e59c82b2dc06ddaa732c01ab343b52683a8f84392080432a7d216bd14829b3d2a762dee2ffe46430c66f298a9b40162d23e9ed5c7183098176a59b037413a","name":"final.geojson","type":"application/geo+json","lastModified":1691073357358.823,"size":183161},{"id":"007304f7665108e003e1af199869f79ba5a8ddd3e761d6bbe70466b6407d258b43b2ea1748af0b0d6d1b38350ff32bfad373a8e99d5df31548bc7fa37666add0","href":"https://static.observableusercontent.com/files/007304f7665108e003e1af199869f79ba5a8ddd3e761d6bbe70466b6407d258b43b2ea1748af0b0d6d1b38350ff32bfad373a8e99d5df31548bc7fa37666add0","name":"final@2.geojson","type":"application/geo+json","lastModified":1691595667723.851,"size":270663}] 2c:{"id":"9847c08c46f56ed6","slug":null,"title":"Overture Places Quality Analysis","thumbnail":"93483d37715016640ac96554c8483a7b904059fe3f626729efb9e9d80603365b","access_level":"public","create_time":"$D2023-08-03T14:35:30.246Z","update_time":"$D2023-08-11T18:56:00.858Z","view_time":null,"first_public_time":"$D2023-08-03T14:35:30.246Z","fork":false,"stars":4,"starred":false,"owner":{"type":"team","id":"524efcbeea8f4655","login":"willemarcel","name":"Wille Marcel","avatar_url":"https://avatars-next.observableusercontent.com/f4de9abe4550086af4d9c3ca9c73673d279b752d094c856a56dd318d6e3e48d7","tier":"starter_2024"},"version":756,"publish_version":null,"role":null,"authored":false,"authors":[{"type":"individual","id":"1ed1697155e35789","login":"willemarcel","name":"Wille Marcel","avatar_url":"https://avatars-next.observableusercontent.com/f4de9abe4550086af4d9c3ca9c73673d279b752d094c856a56dd318d6e3e48d7","tier":null}],"members":[{"type":"individual","id":"1ed1697155e35789","login":"willemarcel","name":"Wille Marcel","avatar_url":"https://avatars-next.observableusercontent.com/f4de9abe4550086af4d9c3ca9c73673d279b752d094c856a56dd318d6e3e48d7","tier":null,"role":"editor"}],"comments":[],"custom_thumbnail":false,"trashed":false,"subscribed":null,"forks":1,"collection":null,"type":"notebook","body":{"title":"Overture Places Quality Analysis","cells":[{"id":0,"value":"# Overture Places Quality Analysis","pinned":false,"hidden":false,"mode":"md"},{"id":204,"value":"Recently, Overture Maps Foundation released a worldwide [Overture Places dataset](https://github.com/OvertureMaps/data), with more than 59 million items. Besides some metadata like name, category, address, website and phone, the items come with a confidence score that varies from 0 to 1, and points out how much Overture is sure the place exists.\n\nIn our analysis, we checked ${features.length} places in a neighbourhood in Salvador de Bahia, Brazil. ","pinned":false,"hidden":false,"mode":"md"},{"id":102,"value":"## Correctness checking","pinned":false,"hidden":false,"mode":"md"},{"id":525,"value":"Each item was checked individually and we learned that ${getPercent(features.filter((i) => i.valid).length / features.length)} of the places are correct, regardless of the confidence score. A correct place means that it is currently active and its coordinates allow users to find it.\n\nHowever, as we show in the next section, the accuracy is very well aligned with the confidence score.","pinned":false,"hidden":false,"mode":"md"},{"id":99,"value":"Plot.plot({\n marginLeft: 60,\n marginRight: 60,\n label: null,\n x: { label: \"Number of places\", domain: [0, 200] },\n y: { padding: 0 },\n marks: [\n Plot.barX(\n features.map((i) => ({\n ...i,\n analysis: i.valid ? \"Correct\" : \"Incorrect\"\n })),\n {\n y: \"analysis\",\n sort: { y: \"x\", order: \"ascending\" },\n x: 1,\n inset: 0.5,\n fill: (d) => (d.valid ? \"steelblue\" : \"#fc8d59\")\n }\n ),\n Plot.ruleX([0])\n ]\n})","pinned":false,"hidden":false,"mode":"ojs"},{"id":9,"value":"$2f","pinned":false,"hidden":false,"mode":"ojs"},{"id":134,"value":"## Accuracy x Confidence score\n\nThe accuracy of the dataset increases with the confidence score. When filtering items with confidence value >= 0.6, the result was ${getPercent(features.filter(i => i.confidence >= 0.6 && i.valid).length / features.filter(i => i.confidence >= 0.6).length)} of accuracy.\n\nMore than that, the accuracy reaches 95% on the places with confidence >= 0.9.","pinned":false,"hidden":false,"mode":"md"},{"id":661,"value":"Plot.plot({\n marks: [\n Plot.ruleY([60]),\n Plot.lineX(confidenceStats, {\n x: \"confidence\",\n y: \"value\",\n stroke: \"steelblue\",\n marker: \"circle\",\n strokeWidth: 2.5\n })\n ],\n grid: true,\n y: { label: \"Percentage of correct places\", domain: [60, 100] },\n x: { label: \"Minimum confidence score\", domain: [0, 1] }\n})","pinned":false,"hidden":false,"mode":"ojs"},{"id":248,"value":"## Categories","pinned":false,"hidden":false,"mode":"md"},{"id":146,"value":"The category values need to be improved to make it easier to use. There are some very similar values, like 'cafe' and 'cafetaria', or 'psychologist' and 'psychotherapist'. Most restaurants have its subcategory (example: 'asian_restaurant') as the main category. The 'landmark_and_historical_building' category has lots of common residential buildings and condominiums.","pinned":false,"hidden":false,"mode":"md"},{"id":59,"value":"Plot.plot({\n facet: { data: features, y: \"category\", marginLeft: 200 },\n marks: [\n Plot.frame({ stroke: \"#ddd\", strokeWidth: 0.5 }),\n Plot.dot(features, { x: \"confidence\", y: \"valid\", fill: \"valid\", r: 3 })\n ],\n width,\n height: 1500,\n marginTop: 0,\n x: { inset: 10, grid: true, label: \"confidence score\" },\n y: { axis: null, inset: 3 },\n color: { legend: true, scheme: \"RdYlBu\" }\n})","pinned":false,"hidden":false,"mode":"ojs"},{"id":286,"value":"## Diving into categories groups","pinned":false,"hidden":false,"mode":"md"},{"id":289,"value":"### Restaurants","pinned":false,"hidden":false,"mode":"md"},{"id":292,"value":"Grouping places that represent restaurants and setting the confidence threshold on 0.6, we got an impressive ${getPercent(21/22)} accuracy.","pinned":false,"hidden":false,"mode":"md"},{"id":256,"value":"DonutChart(restaurantsGroup, {\n name: (d) => d.name,\n value: (d) => d.value,\n innerRadius: 0,\n height: 200,\n colors: [\"steelblue\", \"#fc8d59\"]\n})","pinned":false,"hidden":false,"mode":"ojs"},{"id":305,"value":"### Other groups","pinned":false,"hidden":false,"mode":"md"},{"id":425,"value":"Grouping other places categories and applying the same confidence threshold (0.6), we can see the accuracy of each group.","pinned":false,"hidden":false,"mode":"md"},{"id":356,"value":"Plot.plot({\n marks: [\n Plot.barY(groupsAccuracy, {\n y: \"accuracy\",\n x: \"category\",\n title: (d) => `${d.accuracy.toPrecision(3)}% (${d.items} places)`,\n fill: \"#bfcecd\",\n order: \"fill\",\n sort: { x: \"-y\" }\n }),\n Plot.ruleY([0])\n ],\n x: { label: \"\" },\n y: { label: \"Accuracy (%)\", domain: [0, 100] }\n})","pinned":false,"hidden":false,"mode":"ojs"},{"id":452,"value":"## Data Quality Issues","pinned":false,"hidden":false,"mode":"md"},{"id":491,"value":"For each item in the dataset, we analysed if it has some data quality issues. This section shows the issues found in the correct and in the incorrect places.","pinned":false,"hidden":false,"mode":"md"},{"id":487,"value":"### Correct places","pinned":false,"hidden":false,"mode":"md"},{"id":489,"value":"On the correct places, the main issue is **\"small deviation\"**, what means the place is not perfectly positioned, but, as it's close to the right place, it yet can be found by a map user.\n\nNevertheless, the **\"wrong location\"** means a place that exists in the area of our analysis, but its correct position is on a neighbour street or on another end of an avenue.","pinned":false,"hidden":false,"mode":"md"},{"id":454,"value":"Plot.plot({\n marks: [\n Plot.barY(\n qualityIssues.filter((i) => i.valid),\n Plot.groupX(\n { y: \"count\" },\n {\n x: \"issue\",\n fill: \"steelblue\",\n order: \"fill\",\n sort: { x: \"-y\" }\n }\n ),\n {}\n ),\n Plot.ruleY([0])\n ],\n x: { label: \"\" },\n width,\n y: { label: \"Number of occurrences\", grid: true, domain: [0, 35] }\n})","pinned":false,"hidden":false,"mode":"ojs"},{"id":514,"value":"### Incorrect places\n\nOn the incorrect places, the main issue is **\"outdated\"**, which means the place was active a few months or years ago, but is now closed or moved to another location.\n\nThe **\"wrong location\"** also has many occurrences. On this context, it means that the place exists, but its correct location is not in the analysis area or its surroundings.","pinned":false,"hidden":false,"mode":"md"},{"id":481,"value":"Plot.plot({\n marks: [\n Plot.barY(\n qualityIssues.filter((i) => !i.valid),\n Plot.groupX(\n { y: \"count\" },\n {\n x: \"issue\",\n fill: \"#fc8d59\",\n order: \"fill\",\n sort: { x: \"-y\" }\n }\n ),\n {}\n ),\n Plot.ruleY([0])\n ],\n x: { label: \"\" },\n width,\n y: { label: \"Number of occurrences\", grid: true, domain: [0, 45] }\n})","pinned":false,"hidden":false,"mode":"ojs"},{"id":530,"value":"## Conclusion\n\nThis analysis shows that the Overture Places dataset can be quite useful when filtered by a 0.6 confidence threshold. The main quality issues we observed are the inaccurate location of many items and outdated places. It would also help if we had a better categorization structure.","pinned":false,"hidden":false,"mode":"md"},{"id":629,"value":"----","pinned":false,"hidden":false,"mode":"md"},{"id":442,"value":"qualityIssues = features.reduce((acc, i) => {\n i.quality_issues\n ?.split(\", \")\n .filter((q) => q)\n .forEach((q) =>\n acc.push({ issue: q, valid: i.valid, confidence: i.confidence })\n );\n return acc;\n}, [])","pinned":false,"hidden":false,"mode":"ojs"},{"id":323,"value":"$30","pinned":false,"hidden":false,"mode":"ojs"},{"id":264,"value":"restaurantsGroup = [\n ...d3\n .group(\n features\n .filter((i) => i.category.includes(\"restaurant\"))\n .filter((i) => i.confidence >= 0.6),\n (i) => i.valid\n )\n .entries()\n]\n .map((i) => ({ name: i[0] ? \"Correct\" : \"Incorrect\", value: i[1].length }))\n .sort((a, b) => d3.ascending(a.name, b.name))","pinned":false,"hidden":false,"mode":"ojs"},{"id":311,"value":"function getPercent(v) {\n return `${v.toPrecision(3) * 100}%`;\n}","pinned":false,"hidden":false,"mode":"ojs"},{"id":299,"value":"groups = categoryGroups.map((c) => ({\n category: c.name,\n data: [\n ...d3\n .group(\n features\n .filter((i) => c.categories.includes(i.category))\n .filter((i) => i.confidence >= 0.6),\n (i) => i.valid\n )\n .entries()\n ]\n .map((i) => ({\n name: i[0] ? \"Correct\" : \"Incorrect\",\n value: i[1].length\n }))\n .sort((a, b) => d3.ascending(a.name, b.name))\n}))","pinned":false,"hidden":false,"mode":"ojs"},{"id":543,"value":"metadata = categoryGroups.map((c) => {\n const f = features\n .filter((i) => c.categories.includes(i.category))\n .filter((i) => i.confidence >= 0.6);\n return {\n category: c.name,\n items: f.length,\n website: (f.filter((i) => i.website).length / f.length) * 100,\n phone: (f.filter((i) => i.phone).length / f.length) * 100,\n address:\n (f.filter((i) => Boolean(i.street && Number(i.house_number))).length /\n f.length) *\n 100\n };\n})","pinned":false,"hidden":false,"mode":"ojs"},{"id":587,"value":"metadataGroup = metadata\n .filter((c) => c.items >= 10)\n .reduce((acc, c) => {\n Object.keys(c)\n .filter((k) => ![\"category\", \"items\"].includes(k))\n .forEach((k) =>\n acc.push({ category: c.category, value: c[k], metadata: k })\n );\n return acc;\n }, [])","pinned":false,"hidden":false,"mode":"ojs"},{"id":333,"value":"groupsAccuracy = groups\n .map((g) => ({\n ...g,\n accuracy:\n (g.data[0].value /\n (g.data[0].value + (g.data[1] ? g.data[1].value : 0))) *\n 100,\n items: g.data[0].value + (g.data[1] ? g.data[1].value : 0)\n }))\n .filter((g) => g.items > 9)","pinned":false,"hidden":false,"mode":"ojs"},{"id":6,"value":"data = FileAttachment(\"final@2.geojson\").json()","pinned":false,"hidden":false,"mode":"ojs"},{"id":62,"value":"features = data.features\n .filter((i) => i.properties.valid !== null)\n .map((i) => ({\n ...i.properties,\n confidence: Number(i.properties.confidence)\n }))","pinned":false,"hidden":false,"mode":"ojs"},{"id":647,"value":"confidenceStats\n= [0, 0.1, 0.2, 0.3, 0.4, 0.5, 0.6, 0.7, 0.8, 0.9].map((x) => {\n const p =\n features.filter((i) => i.confidence >= x && i.valid).length /\n features.filter((i) => i.confidence >= x).length;\n return { confidence: x, value: p * 100 };\n})","pinned":false,"hidden":false,"mode":"ojs"},{"id":126,"value":"html``","pinned":false,"hidden":false,"mode":"ojs"},{"id":172,"value":"import { Plot } from \"@mkfreeman/plot-tooltip\"","pinned":true,"hidden":false,"mode":"ojs"},{"id":12,"value":"maplibre = require(\"maplibre-gl@latest\")","pinned":false,"hidden":false,"mode":"ojs"},{"id":16,"value":"xyzStyle = ({\n version: 8,\n glyphs: 'https://fonts.openmaptiles.org/{fontstack}/{range}.pbf',\n sources: {\n 'raster-tiles': {\n type: 'raster',\n tiles: ['https://b.basemaps.cartocdn.com/light_all/{z}/{x}/{y}.png'],\n tileSize: 256,\n attribution:\n '© OpenStreetMap contributors © CARTO'\n }\n },\n layers: [\n {\n id: 'simple-tiles',\n type: 'raster',\n source: 'raster-tiles',\n minzoom: 0,\n maxzoom: 21\n }\n ]\n});\n","pinned":false,"hidden":false,"mode":"ojs"},{"id":258,"value":"import { DonutChart } from \"@d3/donut-chart\"","pinned":true,"hidden":false,"mode":"ojs"}],"files":"$W31","stdlib":"1","titling":"auto"}} 2d:"$2c:body"