31:T112b,md` ## Background Gentrification is a process of displacement and neighborhood transformation that perpetuates inequality in cities. Quantifying gentrification is crucial to policy makers and activists as a way to contain or counter its impacts, but such quantification is paradoxically complex. Macro-level data such as changes in area median income or property values are typically visible after the fact of gentrification, making them useful for understanding the circumstances of neighborhood displacement but doing little to prevent it. Meanwhile, more everyday signifiers of gentrification familiar to anyone living in a city--an excess of [coffee shops](https://www.hbs.edu/faculty/Publication%20Files/18-077_a0e9e3c7-eceb-4685-8d72-21e0f518b3f3.pdf), the creation/upgrading of a [neighborhood dog park](https://nymag.com/intelligencer/2019/03/dog-park-inequality.html), increased use of [modernist sans-serifs](https://www.vice.com/en/article/ep499w/gentrification-font-meme-neutraface)--may be legible on-the-ground to residents, but are less readily translated into legible data for policy purposes. One such on-the-ground but hard-to-quantify gentrification metric anecdotally reported on in New York City is the appearance of expensive/“bougie” products in bodegas. Like the subway, bogedas are a kind of social equalizer in New York: across race, class, and generational divides, the bodega serves all. Bodega owners are [uniquely attuned to neighborhood change](https://www.wnyc.org/story/81216-gentrification-in-brooklyn-one-bodega-at-a-time/) in the form of product requests, which can lead to adding of so-called bougie products to their shelves. While occasionally reported on and memed, bodega inventory change is rarely measured in-depth due to the nuance of such metrics and limited data. But thanks to the emergence of white-label [product locator maps](https://destini.co/), it's possible to track some products' entry and proliferation into bodega-space. ## Problem Statement **Is bodega inventory a meaningful metric of gentrification? Can we quantify the bougie bodega?** We will be exploring this question through inventories of Soylent in New York City. Soylent was chosen for this research in part because Soylent is objectively the funniest thing to see in a bodega; more importantly, however, the company had an online product map and starting in 2018 an aggressive retail strategy strengthened in New York City by their [partnership with Big Geyser](https://www.prnewswire.com/news-releases/soylent-secures-major-new-york-retail-distribution-partnership-with-big-geyser-300592438.html), the largest independent beverage distributor in the NYC metro area. Data collection on Soylent took place in early 2020 (pre-pandemic, when silly side projects seemed feasible) and late 2021 (post-Delta but pre-Omicron, when I needed to come up with a final project topic and it had to include some time series analysis). ### Limitations/Caveats Asking a New Yorker if some place is "really" a bodega is a known way to start a fight. (Anecdotal surveying of peers emphasizes the presence of a cat and the ability to buy both cigarettes and breakfast in the same place; your mileage may vary) My process for identifying bodegas was manually reviewing the data to identify places that were obviously (to me) not bodegas (basically: CVS, 7-11, gas stations, fancy gyms with cafes). Maybe the important metric here isn't merely "bodega", but places where inventory is decided by a local small business owner rather than a corporate higher-up. Another factor to acknowledge here is the reality of how distribution companies actually work with bodegas and other small businesses. Often, when trying to sell a new product distributors will make deals with business owners to convince them to take the new product--for example, in exchange for taking a case of new product X to see how it does, the distributor will offer a discount or additional inventory of the more reliable product Y that the business owner would be buying anyway. Choosing to sell Soylent is not always necessarily a consequence of local demand. We also don't have a full count of all bodegas (or all CVSes/7-11s etc.) to grasp the scale of Soylent availability relative to all bodegas in the city, so our grasp of the product's penetration of retail establishments it s a little unclear. `32:T9d5,soylentmap = { let bodegastyle = { radius: 4, color: "white", fillColor: "rebeccapurple", weight: 1, opacity: 1, fillOpacity: 0.8 }; let notbodegastyle = { radius: 4, color: "white", fillColor: "teal", weight: 1, opacity: 1, fillOpacity: 0.8 }; let soylent21style = { radius: 4, color: "white", fillColor: "darkgoldenrod", weight: 1, opacity: 1, fillOpacity: 0.8 }; let bodegas = L.geoJSON( turf.featureCollection( soylent20.features.filter( (f) => f.properties.bodega_likely == "Possible Bodega" ) ), { pointToLayer: function (feature, latlng) { return L.circleMarker(latlng, bodegastyle); }, onEachFeature: onEachFeature } ); let notbodegas = L.geoJSON( turf.featureCollection( soylent20.features.filter( (f) => f.properties.bodega_likely == "Definitely Not a Bodega" ) ), { pointToLayer: function (feature, latlng) { return L.circleMarker(latlng, notbodegastyle); }, onEachFeature: onEachFeature } ); let all21 = L.geoJSON(soylent21, { pointToLayer: function (feature, latlng) { return L.circleMarker(latlng, soylent21style); }, onEachFeature: onEachFeature }); //added height to the div let container = DOM.element("div", { style: "height:800px;" }); yield container; let map = L.map(container, { layers: [bodegas, notbodegas, all21] }).setView( [40.6829, -73.9708], 11 ); let overlayMaps = { "Possible Bodegas (2020)": bodegas, "Not Bodegas (2020)": notbodegas, "Soylent Vendors (2021)": all21 }; L.control.layers("", overlayMaps, { collapsed: false }).addTo(map); let osmLayer = L.tileLayer( "https://tiles.stadiamaps.com/tiles/stamen_toner_lite/{z}/{x}/{y}{r}.png", { attribution: '© OpenStreetMap contributors, Stadia Maps, and Stamen' } ).addTo(map); function onEachFeature(feature, layer) { // does this feature have a property named popupContent? if (feature.properties && feature.properties.name) { layer.bindPopup( `${feature.properties.name}
${feature.properties.address}` ); } } }33:T771,So it's pretty obvious from this that Soylent vendors overall were concentrated in Manhattan and North Brooklyn in 2020 (which is maybe not especially surprising?). The distribution of bodega to not-bodega in 2020 is actually pretty close: around 51% of the vendors here are classified as possible bodegas, and given the likelihood of error in some of the categorizing it wouldn't surprise me if it's more like 50/50 or even 49/51 the other way. (One detail of having an interactive map of this is it makes picking out not-bodegas easier!) It's also clear that 2021 saw a big drop-off in Soylent vendors. Between 2020 and 2021, Soylent availability dropped overall by 77.9% across all of New York City--from 1,125 vendors to 248. A number of external factors likely shaped the state of Soylent retail between 2020 and 2021. The COVID-19 pandemic led to many bodegas [closing up shop](https://ny.eater.com/2020/4/10/21214971/nyc-bodegas-coronavirus-survival-grocery), which presumably reduced the number of Soylent vendors. But the most significant factor would be the apparent termination of Big Geyser's relationship with Soylent sometime in 2020. As the largest independent non-alcoholic beverage distributor in the New York metro area, Big Geyser would have been primarily responsible for convincing bodega owners to carry Soylent in the first place. This is why there's no differentiation between bodega or non-bodega for the 2021 data. The main NYC Soylent vendors left are chains like CVS and Target. Another factor to weigh here is [Soylent's business strategy](https://fortune.com/2021/07/28/soylent-retail-nutrition-rite-aid-walmart/): the company is courting tech elites less and less, and increasingly seems to be focused on different niche audiences like busy parents. It's possible that Soylent wasn't seeing particularly good retail sales in New York and opted to focus on other markets. 34:T63a,udpnycmap = { let udp = L.geoJSON(udpnyc, { style: categoricalStyle, onEachFeature: onEachFeature }); //added height to the div let container = DOM.element("div", { style: "height:800px;" }); yield container; let map = L.map(container, { layers: [udp] }).setView( [40.6829, -73.9708], 11 ); let overlayMaps = { "All Typologies": udp }; function categoricalStyle(feature) { return { fillColor: categoricalcolor(feature.properties.typology_origin), weight: 0.5, opacity: 1, color: "white", fillOpacity: 0.7 }; } // L.control.layers(overlayMaps, { collapsed: false }).addTo(map); let osmLayer = L.tileLayer( "https://tiles.stadiamaps.com/tiles/stamen_toner_lite/{z}/{x}/{y}{r}.png", { attribution: '© OpenStreetMap contributors, Stadia Maps, and Stamen' } ).addTo(map); function onEachFeature(feature, layer) { if (feature.properties) { layer.bindPopup( `Type: ${feature.properties.typology_origin}
Neighborhood: ${feature.properties.municipali}` ); } } var legend = L.control({ position: "bottomright" }); legend.onAdd = function (map) { var div = L.DomUtil.create("div", "info legend"); let html = buildMap(types, colorramp) .map((m) => ` ${m[0]}
`) .join(""); div.innerHTML = html; return div; }; legend.addTo(map); }35:T9fd,onlysoylentmap = { let colors2020 = d3.quantize(d3.interpolateViridis, 5); let udp20 = L.geoJSON(udpnyc_withsoylent_20, { style: choropleth20, onEachFeature: onEachFeature }); let udp21 = L.geoJSON(udpnyc_withsoylent_21, { style: choropleth21, onEachFeature: onEachFeature }); //added height to the div let container = DOM.element("div", { style: "height:800px;" }); yield container; let map = L.map(container, { layers: [udp20] }).setView( [40.6829, -73.9708], 11 ); let overlayMaps = { "Soylent Vendors (2020)": udp20, "Soylent Vendors (2021)": udp21 }; function categoricalStyle(feature) { return { fillColor: categoricalcolor(feature.properties.typology_origin), weight: 0.5, opacity: 1, color: "white", fillOpacity: 0.7 }; } function choropleth20(feature, array) { return { fillColor: getColor(colors2020, feature.properties.soylent2020), weight: 0.5, opacity: 1, color: "white", fillOpacity: 0.7 }; } function choropleth21(feature, array) { return { fillColor: getColor(colors2020, feature.properties.soylent2021), weight: 0.5, opacity: 1, color: "white", fillOpacity: 0.7 }; } function getColor(arr, d) { let a = arr.filter((c, i) => d == i + 1); if (a.length == 0) { return "#fde725"; } else { return a.toString(); } } L.control.layers(overlayMaps, "", { collapsed: false }).addTo(map); let osmLayer = L.tileLayer( "https://tiles.stadiamaps.com/tiles/stamen_toner_lite/{z}/{x}/{y}{r}.png", { attribution: '© OpenStreetMap contributors, Stadia Maps, and Stamen' } ).addTo(map); function onEachFeature(feature, layer) { if (feature.properties) { layer.bindPopup( `${feature.properties.typology_origin}
${feature.properties.municipali}
Soylent Vendors 2020: ${feature.properties.soylent2020}
Soylent Vendors 2021: ${feature.properties.soylent2021}` ); } } var legend = L.control({ position: "bottomright" }); legend.onAdd = function (map) { var div = L.DomUtil.create("div", "info legend"); let html = colors2020 .map((m, i) => `
${i + 1}`) .join(""); div.innerHTML = `Soylent Vendor Count${html}+`; return div; }; legend.addTo(map); }36:T40f,tractplot = Plot.plot({ width: width, x: { domain: [20, 45], label: "Percent tracts with Soylent vendors (2020)", percent: true }, y: { // domain: [0, 25], label: "Percent tracts with Soylent vendors (2021)", percent: true }, marks: [ Plot.line( udpnyc_by_type, RegPlot.regression({ x: "percent_soylent_20", y: "percent_soylent_21", strokeWidth: 2, stroke: "turquoise", strokeOpacity: 0.5 }) ), Plot.line( udpnyc_by_type_nooutlier, RegPlot.regression({ x: "percent_soylent_20", y: "percent_soylent_21", strokeWidth: 2, stroke: "mediumpurple", strokeOpacity: 0.5 }) ), Plot.text(udpnyc_by_type, { x: "percent_soylent_20", y: "percent_soylent_21", text: "abbrev", dx: -5, // dy: 3, textAnchor: "end" }), Plot.dot(udpnyc_by_type, { x: "percent_soylent_20", y: "percent_soylent_21", fill: "gray" }) ] })37:T463,## Next steps/further research While the correlation in this analysis is pretty strong for gentrification status and Soylent availability both with and without bodega vendors, given we're actually only dealing with 7 data points I'd still remain cautious about the findings here. Also, I hate the findings and thus I do not want to take them too seriously. This research does suggest that bodega products are a promising avenue for understanding gentrification's spread and might be compelling to replicate with other products currently entering bodega retail (CBD beverage [Recess](https://takearecess.com/where-to-buy) seems promising, as does [VYBES](https://idrinkvybes.com/pages/store-locator)). Another wrinkle of this research, of course, is we have little grasp of bodega saturation as we don't have a count of all bodegas in New York City. Jeff Sisson's 2009 [Bodega List](https://rhizome.org/editorial/2009/jan/20/bodega-list-2009-jeff-sisson/) project offers a promising methodology; consulting with bodega experts such as [The Bodega Boys](https://www.sho.com/desus-and-mero) is a longer-term aspiration. ***38:T534,udpnyc_by_type = types.map((t) => { let obj = { type: t, total: udpnyc.features.filter((u) => u.properties.typology_origin == t) .length, tracts_20: udpnyc_withsoylent_20.filter( (u) => u.properties.typology_origin == t ).length, tracts_21: udpnyc_withsoylent_21.filter( (u) => u.properties.typology_origin == t ).length, vendors_20: udpnyc_withsoylent_20 .filter((u) => u.properties.typology_origin == t) .map((m) => m.properties.soylent2020) .reduce((a, b) => a + b, 0), bodega_20: udpnyc_withsoylent_20 .filter((u) => u.properties.typology_origin == t) .map((m) => m.properties.soylent2020_bodega) .reduce((a, b) => a + b, 0), vendors_21: udpnyc_withsoylent_21 .filter((u) => u.properties.typology_origin == t) .map((m) => m.properties.soylent2021) .reduce((a, b) => a + b, 0) }; obj.change_tracts = 1 - obj.tracts_21 / obj.tracts_20; obj.change_vendors = 1 - obj.vendors_21 / obj.vendors_20; obj.percent_soylent_20 = obj.tracts_20 / obj.total; obj.percent_soylent_21 = obj.tracts_21 / obj.total; obj.abbrev = obj.type.split("-").filter((f) => f != "At")[0] + obj.type .split(" ") .filter((f) => f != "At") .filter((f) => f != "Highly")[2] .replace(",", ""); return obj; })39:T446,// borrowed from @Fil but we only really need linear regression, this isn't especially good science lol function addRegression(Plot) { Plot.regression = function ({ x, y, type, bandwidth, order, ...options }) { type = String(type).toLowerCase(); const regressor = reg.regressionLinear(); const z = options.z || options.stroke; // maybeZ return Plot.transform(options, function (data, facets) { const X = Plot.valueof(data, x); const Y = Plot.valueof(data, y); const Z = Plot.valueof(data, z); regressor.x((i) => X[i]).y((i) => Y[i]); const regFacets = []; const points = []; for (const facet of facets) { const regFacet = []; for (const I of Z ? d3.group(facet, (i) => Z[i]).values() : [facet]) { const reg = regressor(I); for (const d of reg) { const j = points.push(d) - 1; if (z) d[z] = Z[I[0]]; regFacet.push(j); } } regFacets.push(regFacet); } return { data: points, facets: regFacets }; }); }; return Plot; }3a:[{"id":"3072a938fe154f03f16ec23a4515cf6751cd244145a65696b60a4e2b4e09bdad7244b36548612638e5894e98177925c66ec34672d33e1143ec87ee0e52f33c4a","href":"https://static.observableusercontent.com/files/3072a938fe154f03f16ec23a4515cf6751cd244145a65696b60a4e2b4e09bdad7244b36548612638e5894e98177925c66ec34672d33e1143ec87ee0e52f33c4a","name":"soylent-2021.geojson","type":"application/geo+json","lastModified":1639681458673.364,"size":41352},{"id":"164fdac3ea896db4a83d8dcd9641d96e29fc6f69b065fd1917d1e58c70e2d9e0445001a92917c503c24cbaed570a0502542e8dcce9100e7bdc9f04121c1e5bc6","href":"https://static.observableusercontent.com/files/164fdac3ea896db4a83d8dcd9641d96e29fc6f69b065fd1917d1e58c70e2d9e0445001a92917c503c24cbaed570a0502542e8dcce9100e7bdc9f04121c1e5bc6","name":"soylent2020-categorized.geojson","type":"application/geo+json","lastModified":1639933250942.271,"size":261954},{"id":"0ccd30c4cab99d4bf542e656c6fa55c12d03f5273ce313c5dd9530525f81d678a406f78d3869c9ae19b85d0a52351392efa7c0ff2506b7e13bfdb341f8da7489","href":"https://static.observableusercontent.com/files/0ccd30c4cab99d4bf542e656c6fa55c12d03f5273ce313c5dd9530525f81d678a406f78d3869c9ae19b85d0a52351392efa7c0ff2506b7e13bfdb341f8da7489","name":"udpnyc-soylentcounts.geojson","type":"application/geo+json","lastModified":1639933256096.142,"size":1559969}] 2f:{"title":"Bodega Inventory as Gentrification Metric","cells":[{"id":0,"value":"# Bodega Inventory as Gentrification Metric","pinned":false,"hidden":false,"mode":"md"},{"id":919,"value":"This is an adaptation of a final project I did for a data science class, originally done in a Jupyter notebook. I regret everything that follows. ","pinned":false,"hidden":false,"mode":"md"},{"id":50,"value":"$31","pinned":false,"hidden":false,"mode":"ojs"},{"id":55,"value":"md`## Datasets\n\n- **soylent20**: data scraped from Soylent product map and geocoded, January 2020 \n- **soylent21**: data scraped from Soylent product map and geocoded, November 2021\n- **udpnyc**: data scraped from Carto visualization of 2018 [Urban Displacement Project NY Metro Region](http://udpny.org/) project`","pinned":false,"hidden":false,"mode":"ojs"},{"id":209,"value":"## Let's look at the data","pinned":false,"hidden":false,"mode":"md"},{"id":194,"value":"### All Soylent Vendors, 2020 and 2021","pinned":false,"hidden":false,"mode":"md"},{"id":104,"value":"$32","pinned":false,"hidden":false,"mode":"ojs"},{"id":298,"value":"$33","pinned":false,"hidden":false,"mode":"md"},{"id":316,"value":"### Soylent Vendor Distribution Across Urban Displacement Typologies","pinned":false,"hidden":false,"mode":"md"},{"id":1059,"value":"#### First: a general look at the Urban Displacement Typology Map","pinned":false,"hidden":false,"mode":"md"},{"id":718,"value":"$34","pinned":false,"hidden":false,"mode":"ojs"},{"id":1163,"value":"I don't want to rewrite the entirety of the source's methodology here, but just to clarify the typologies: LI = Low Income, MHI = Middle to High Income, VLI = Very Low Income. So the types here are broken out not just by income but by risk of gentrification/displacement and in middle-to-high income areas, exclusion (i.e., the area has displaced low-income residents and there are significant barriers for low-income residents to move here). ","pinned":false,"hidden":false,"mode":"md"},{"id":666,"value":"#### Total number of Soylent vendors per displacement typology, 2020","pinned":false,"hidden":false,"mode":"md"},{"id":612,"value":"Plot.legend({\n color: {\n range: [\"mediumpurple\", \"lightgray\"],\n domain: [\"Total\", \"Possible Bodega\"]\n },\n legend: \"swatches\",\n swatchSize: 10,\n style: {\n fontSize: \"15px\"\n }\n})","pinned":false,"hidden":false,"mode":"ojs"},{"id":276,"value":"Plot.plot({\n x: { label: \"Displacement Typology\" },\n y: { label: \"Soylent Vendors (2020)\" },\n width: width,\n marks: [\n Plot.barY(udpnyc_by_type, {\n x: \"abbrev\",\n y: \"vendors_20\",\n fill: \"mediumpurple\"\n }),\n Plot.barY(udpnyc_by_type, {\n x: \"abbrev\",\n y: \"bodega_20\",\n fill: \"lightgray\"\n }),\n Plot.text(udpnyc_by_type, {\n x: \"abbrev\",\n y: \"bodega_20\",\n text: \"bodega_20\",\n dy: -5,\n fill: \"white\"\n }),\n Plot.text(udpnyc_by_type, {\n x: \"abbrev\",\n y: \"vendors_20\",\n text: \"vendors_20\",\n dy: -5\n })\n ]\n})","pinned":false,"hidden":false,"mode":"ojs"},{"id":676,"value":"At an aggregate level, areas of New York City classified as low-income and at risk of gentification and/or displacement, middle-to-high income already at an advanced stage of gentrification, middle-to-high income and stable, or middle-to-high income and at risk of exclusion had the most Soylent vendors in 2020. The drop-off in Soylent vendors between 2020 and 2021 is also pretty visible with the average decrease (75.5%) closely aligned with the citywide loss (77.9%).","pinned":false,"hidden":false,"mode":"md"},{"id":670,"value":"#### Total Soylent vendors per displacement typology, 2020 and 2021","pinned":false,"hidden":false,"mode":"md"},{"id":649,"value":"Plot.legend({\n color: {\n range: [\"mediumpurple\", \"teal\"],\n domain: [\"2020\", \"2021\"]\n },\n legend: \"swatches\",\n swatchSize: 10,\n style: {\n fontSize: \"15px\"\n }\n})","pinned":false,"hidden":false,"mode":"ojs"},{"id":643,"value":"Plot.plot({\n x: { label: \"Displacement Typology\" },\n y: { label: \"Soylent Vendors\" },\n width: width,\n marks: [\n Plot.barY(udpnyc_by_type, {\n x: \"abbrev\",\n y: \"vendors_20\",\n fill: \"mediumpurple\"\n }),\n Plot.barY(udpnyc_by_type, {\n x: \"abbrev\",\n y: \"vendors_21\",\n fill: \"teal\"\n }),\n Plot.text(udpnyc_by_type, {\n x: \"abbrev\",\n y: \"vendors_21\",\n text: \"vendors_21\",\n dy: -5,\n fill: \"white\"\n }),\n Plot.text(udpnyc_by_type, {\n x: \"abbrev\",\n y: \"vendors_20\",\n text: \"vendors_20\",\n dy: -5\n })\n ]\n})","pinned":false,"hidden":false,"mode":"ojs"},{"id":714,"value":"However, in early 2020 only 29.3% of _all_ New York City census tracts had _any_ stores selling Soylent--overall penetration wasn't all that high to begin with, and it dropped to 10.7% in 2021. As seen in the 2020 maps, the heaviest concentrations of Soylent vendors were in Manhattan and North Brooklyn. Once we remove the areas untouched by Soylent, the data gets a little more interesting. ","pinned":false,"hidden":false,"mode":"md"},{"id":719,"value":"Looking at the _percentage_ of census tracts within each typology that had any Soylent vendors, the distribution of Soylent availability isn't quite as disparate as the simple counts of vendors. The areas with high Soylent counts (middle-to-high income and at advanced gentrification, middle-to-high income and at risk of exclusion, and low-income areas subject to ongoing gentrification) have the highest Soylent representation across their census tracts. ","pinned":false,"hidden":false,"mode":"md"},{"id":1159,"value":"#### Census tracts with at least one Soylent vendor, 2020 and 2021","pinned":false,"hidden":false,"mode":"md"},{"id":1086,"value":"$35","pinned":false,"hidden":false,"mode":"ojs"},{"id":851,"value":"#### Percent of census tracts with any Soylent vendors, 2020 and 2021","pinned":false,"hidden":false,"mode":"md"},{"id":723,"value":"Plot.legend({\n color: {\n range: [\"mediumpurple\", \"teal\"],\n domain: [\"2020\", \"2021\"]\n },\n legend: \"swatches\",\n swatchSize: 10,\n style: {\n fontSize: \"15px\"\n }\n})","pinned":false,"hidden":false,"mode":"ojs"},{"id":278,"value":"percentsplot = Plot.plot({\n x: { label: \"Displacement Typology\" },\n y: { label: \"Percent tracts with Soylent vendors (2020)\", percent: true },\n width: width,\n marks: [\n Plot.barY(udpnyc_by_type, {\n x: \"abbrev\",\n y: \"percent_soylent_20\",\n fill: \"mediumpurple\"\n }),\n Plot.barY(udpnyc_by_type, {\n x: \"abbrev\",\n y: \"percent_soylent_21\",\n fill: \"teal\"\n }),\n Plot.text(udpnyc_by_type, {\n x: \"abbrev\",\n y: \"percent_soylent_21\",\n text: (d) => `${(d.percent_soylent_21 * 100).toFixed(1)}%`,\n dy: -5,\n fill: \"white\"\n }),\n Plot.text(udpnyc_by_type, {\n x: \"abbrev\",\n y: \"percent_soylent_20\",\n text: (d) => `${(d.percent_soylent_20 * 100).toFixed(1)}%`,\n dy: -5\n })\n ]\n})","pinned":false,"hidden":false,"mode":"ojs"},{"id":765,"value":"### I regret to inform you that I looked for correlations\n\nI'm hesitant to treat this as an especially impressive finding considering this is only dealing with 8 data points, but there are _sort of_ correlations between changes in Soylent availability between 2020-21 both by tracts and by total count. It does look like despite the steep decline in Soylent availability overall, typologies that had higher Soylent availability than other typologies in 2020 were likely to still have higher Soylent availability in 2021. ","pinned":false,"hidden":false,"mode":"md"},{"id":857,"value":"#### Correlation between total Soylent vendors in 2020 and 2021?","pinned":false,"hidden":false,"mode":"md"},{"id":804,"value":"Plot.legend({\n color: {\n range: [\"mediumpurple\"],\n domain: [\"r-squared .89\"]\n },\n legend: \"swatches\",\n swatchSize: 10,\n style: {\n fontSize: \"15px\"\n }\n})","pinned":false,"hidden":false,"mode":"ojs"},{"id":274,"value":"vendorplot = Plot.plot({\n width: width,\n x: {\n domain: [0, 220],\n label: \"Total Soylent vendors (2020)\"\n },\n y: {\n // domain: [0, 60],\n label: \"Total Soylent vendors (2021)\"\n },\n marks: [\n Plot.line(\n udpnyc_by_type,\n RegPlot.regression({\n x: \"vendors_20\",\n y: \"vendors_21\",\n strokeWidth: 2,\n stroke: \"mediumpurple\",\n strokeOpacity: 0.5\n })\n ),\n Plot.text(udpnyc_by_type, {\n x: \"vendors_20\",\n y: \"vendors_21\",\n text: \"abbrev\",\n dx: -5,\n // dy: 3,\n textAnchor: \"end\"\n }),\n Plot.dot(udpnyc_by_type, {\n x: \"vendors_20\",\n y: \"vendors_21\",\n fill: \"gray\"\n })\n ]\n})","pinned":false,"hidden":false,"mode":"ojs"},{"id":815,"value":"There is one significant outlier data point when looking at change in percentage of tracts with any Soylent availability, which are the middle-to-high income exclusionary tracts. This typology also had the fewest Soylent vendors overall and the lowest percentage of bodega vendors in 2020, which means that it probably retained a lot of its 2020 Soylent vendors. If we treat these conditions as an outlier, the r-squared is...better than I would like because this was sort of supposed to be a dumb little troll of a project. ","pinned":false,"hidden":false,"mode":"md"},{"id":862,"value":"#### Correlation between percent of tracts with Soylent availability in 2020 and 2021?","pinned":false,"hidden":false,"mode":"md"},{"id":783,"value":"Plot.legend({\n color: {\n range: [\"mediumpurple\", \"turquoise\"],\n domain: [\"r-squared .28\", \"r-squared .92 (without MHI Exclusion)\"]\n },\n legend: \"swatches\",\n swatchSize: 10,\n opacity: 0.5,\n style: {\n fontSize: \"15px\"\n }\n})","pinned":false,"hidden":false,"mode":"ojs"},{"id":272,"value":"$36","pinned":false,"hidden":false,"mode":"ojs"},{"id":867,"value":"The bottom line, apparently: \n\n1. Middle-to-high income areas already deep in gentrification or at risk of becoming exclusionary and low-income areas subject to ongoing gentrification tended to have more Soylent availability than areas not experiencing displacement of low-income residents or low-income areas considered at risk of displacement.\n2. Admittedly, though, \"more Soylent availability\" wasn't a matter of like significant difference--the typology with the highest percentage of tracts with Soylent vendors only had 43% coverage in 2020 and that dropped to 16% in 2021.\n4. Bodega availability may be an important factor here; on average most typologies were within the 50% bodega to not-bodega split in 2020; areas with fewer bodega Soylent vendors in 2020 experienced slightly less dramatic vendor losses in 2021.\n5. I hate everything about this.","pinned":false,"hidden":false,"mode":"md"},{"id":327,"value":"$37","pinned":false,"hidden":false,"mode":"md"},{"id":835,"value":"# Appendix","pinned":false,"hidden":false,"mode":"md"},{"id":1234,"value":"## Other Fun Supplemental Stuff","pinned":false,"hidden":false,"mode":"md"},{"id":1239,"value":"### Census tracts with more than 5 Soylent vendors, by UDP type","pinned":false,"hidden":false,"mode":"md"},{"id":1202,"value":"Inputs.table(toptracts)","pinned":true,"hidden":false,"mode":"ojs"},{"id":1259,"value":"### Top Soylent tracts (2020)","pinned":false,"hidden":false,"mode":"md"},{"id":1257,"value":"Inputs.table(tops)","pinned":true,"hidden":false,"mode":"ojs"},{"id":1261,"value":"I do think it's interesting that the UDPNYC data treats Hudson Yards, Chelsea, Flatiron, and Union Square as a single \"neighborhood\"? Also damn, West Village. ","pinned":false,"hidden":false,"mode":"md"},{"id":1230,"value":"## Maths on UDP NYC data","pinned":false,"hidden":false,"mode":"md"},{"id":219,"value":"udpnyc_withsoylent_20 = udpnyc.features.filter(\n (u) => u.properties.soylent2020 > 0\n)","pinned":false,"hidden":false,"mode":"ojs"},{"id":222,"value":"udpnyc_withsoylent_21 = udpnyc.features.filter(\n (u) => u.properties.soylent2021 > 0\n)","pinned":false,"hidden":false,"mode":"ojs"},{"id":225,"value":"types = [...new Set(udpnyc.features.map((u) => u.properties.typology_origin))]","pinned":false,"hidden":false,"mode":"ojs"},{"id":230,"value":"$38","pinned":false,"hidden":false,"mode":"ojs"},{"id":823,"value":"Inputs.table(udpnyc_by_type)","pinned":true,"hidden":false,"mode":"ojs"},{"id":1179,"value":"highest = udpnyc.features.filter((f) => f.properties.soylent2020 > 5)","pinned":true,"hidden":false,"mode":"ojs"},{"id":1213,"value":"mhi_a_tops = highest\n .filter(\n (h) => h.properties.typology_origin == \"MHI - Advanced Gentrification\"\n )\n .map((m) => m.properties)","pinned":true,"hidden":false,"mode":"ojs"},{"id":1217,"value":"mhi_r_tops = highest\n .filter((h) => h.properties.typology_origin == \"MHI - At Risk of Exclusion\")\n .map((m) => m.properties)","pinned":true,"hidden":false,"mode":"ojs"},{"id":1248,"value":"tops = mhi_a_tops.concat(mhi_r_tops).sort((s) => s.soylent20).map(m => {return {\n type: m.typology_origin,\n soylent2020: m.soylent2020,\n bodega2020: m.soylent2020_bodega,\n soylent2021: m.soylent2021,\n neighborhood: m.municipali\n};})","pinned":true,"hidden":false,"mode":"ojs"},{"id":1185,"value":"toptracts = types\n .map((t) => {\n let subset = highest.filter((u) => u.properties.typology_origin == t);\n let obj = {\n type: t,\n \"tracts with 5+ soylent vendors\": subset.length,\n \"total in high-volume areas\": subset\n .map((m) => m.properties.soylent2020)\n .reduce((a, b) => a + b, 0)\n };\n return obj;\n })\n .sort()","pinned":false,"hidden":false,"mode":"ojs"},{"id":687,"value":"average = (array) => array.reduce((a, b) => a + b) / array.length","pinned":true,"hidden":false,"mode":"ojs"},{"id":439,"value":"vendorRegression = reg\n .regressionLinear()\n .x((d) => d.vendors_20)\n .y((d) => d.vendors_21)","pinned":false,"hidden":false,"mode":"ojs"},{"id":442,"value":"vendorRegression(udpnyc_by_type)","pinned":true,"hidden":false,"mode":"ojs"},{"id":793,"value":"percentsRegression = reg\n .regressionLinear()\n .x((d) => d.percent_soylent_20)\n .y((d) => d.percent_soylent_21)","pinned":true,"hidden":false,"mode":"ojs"},{"id":795,"value":"percentsRegression(udpnyc_by_type_nooutlier)","pinned":true,"hidden":false,"mode":"ojs"},{"id":798,"value":"percentsRegression(udpnyc_by_type)","pinned":true,"hidden":false,"mode":"ojs"},{"id":356,"value":"udpnyc_by_type_nooutlier = udpnyc_by_type.filter(\n (u) => u.type != \"MHI - Exclusion\"\n)","pinned":false,"hidden":false,"mode":"ojs"},{"id":1073,"value":"## Map Styling Things","pinned":false,"hidden":false,"mode":"md"},{"id":1095,"value":"colors2020 = d3.quantize(d3.interpolateViridis, 6)","pinned":true,"hidden":false,"mode":"ojs"},{"id":1049,"value":"","pinned":false,"hidden":false,"mode":"html"},{"id":930,"value":"colorramp = d3.quantize(d3.interpolateViridis, 8)","pinned":false,"hidden":false,"mode":"ojs"},{"id":962,"value":"function categoricalcolor(d) {\n let c = buildMap(types, colorramp).filter((b) => b[0] == d);\n return c[0][1];\n}","pinned":false,"hidden":false,"mode":"ojs"},{"id":992,"value":"buildMap = (keys, values) => {\n const map = new Map();\n for (let i = 0; i < keys.length; i++) {\n map.set(keys[i], values[i]);\n }\n return Array.from(map);\n}","pinned":false,"hidden":false,"mode":"ojs"},{"id":1093,"value":"getColor = (arr, d) => {\n let a = arr.filter((c, i) => d == i + 1);\n if (a.length == 0) {\n return \"#fde725\";\n } else {\n return a.toString();\n }\n}","pinned":false,"hidden":false,"mode":"ojs"},{"id":77,"value":"md`## Libraries`","pinned":false,"hidden":false,"mode":"ojs"},{"id":16,"value":"d3 = require(\"d3\")","pinned":false,"hidden":false,"mode":"ojs"},{"id":72,"value":"turf = require(\"@turf/turf\")","pinned":false,"hidden":false,"mode":"ojs"},{"id":100,"value":"L = require(\"leaflet@1.2.0\")","pinned":true,"hidden":false,"mode":"ojs"},{"id":102,"value":"html``","pinned":false,"hidden":false,"mode":"ojs"},{"id":343,"value":"RegPlot = addRegression(Plot_)","pinned":true,"hidden":false,"mode":"ojs"},{"id":516,"value":"$39","pinned":false,"hidden":false,"mode":"ojs"},{"id":349,"value":"reg = require(\"d3-regression@1\")","pinned":true,"hidden":false,"mode":"ojs"},{"id":341,"value":"import { Plot as Plot_ } from \"@fil/plot-splom\"","pinned":true,"hidden":false,"mode":"ojs"},{"id":172,"value":"## Data","pinned":false,"hidden":false,"mode":"md"},{"id":7,"value":"udpnyc = FileAttachment(\"udpnyc-soylentcounts.geojson\").json()","pinned":true,"hidden":false,"mode":"ojs"},{"id":33,"value":"soylent20 = FileAttachment(\"soylent2020-categorized.geojson\").json()","pinned":true,"hidden":false,"mode":"ojs"},{"id":39,"value":"soylent21 = FileAttachment(\"soylent-2021.geojson\").json()","pinned":true,"hidden":false,"mode":"ojs"}],"files":"$W3a","stdlib":"1","titling":"auto"}