Steven Ponce

@sponce1

Website: GitHub: LinkedIn:

883 Following646 Followers402 Posts

Bluesky

Is this your profile?Claim it and sponce1.dirty.social is yours.Claim profile

Steven Ponce@sponce1· 8d

📊 #MakeoverMonday – 2026 W31 | America’s Services Trade Balances with Its Free Trade Partners . 🔗: stevenponce.netlify.app/data_visuali... . #rstats | #DataFam | #dataviz | #ggplot2

Diverging column chart comparing the U.S. services trade balance to the overall (goods + services) trade balance for three countries in 2024. Canada: services +$34.9B, overall −$35.7B. South Korea: services +$10.7B, overall −$55.5B. Mexico: services +$2.5B, overall −$179.0B. All three show a positive services balance but a negative overall balance — the only 3 of 15 U.S. free-trade partners where this pattern holds in 2024.

Steven Ponce@sponce1· 10d

📊 #SWDchallenge Aug2026 | whip up a waterfall . FDA review approves 9 in 10 drugs that reach it — the real bottleneck is Phase II, where 71% of candidates never advance. . 🔗 stevenponce.netlify.app/data_visuali... . #dataviz | #rstats | #DataStorytelling

Waterfall chart titled "The stage with the lowest odds isn't FDA review." The chart tracks a hypothetical cohort of 1,000 experimental drugs from Phase I trials through FDA approval, with floating bars showing how many candidates are lost at each development stage. Phase II proof-of-concept testing, highlighted in burgundy, eliminates more candidates than any other single stage — 370, compared to 480 in Phase I safety testing, 63 in Phase III confirmation, and just 8 during FDA review. Only 28.9% of candidates advance beyond Phase II, versus a 90.6% approval rate at FDA review. The bridge ends with 79 of the original 1,000 candidates reaching approval. Source: BIO, Informa Pharma Intelligence and QLS Advisors, Clinical Development Success Rates 2011–2020.

Steven Ponce@sponce1· 11d

📊 #TidyTuesday – 2026 W31 | Basotho Wool . 🔗: stevenponce.netlify.app/data_visuali... . #rstats | #r4ds | #dataviz | #ggplot2

Two-panel bar chart, "One Producer, Two Calendars." Each bar shows a month's trade value as a percent deviation from that importer's own average, 2016–2024. Top panel: South Africa's Basotho wool imports peak in October–November (about +150%) and trough April–July (down to -93%). Bottom panel: China's imports peak in April (+133%) and trough August–October (down to -88%, based on only 4–6 annual observations). The same wool follows opposite seasonal calendars depending on destination. Data are partner-reported imports from Lesotho via UN Comtrade, not export or shearing records.

Steven Ponce@sponce1· 15d

📊 #MakeoverMonday – 2026 W30 | Squirrel Census 2020 . 🔗: stevenponce.netlify.app/data_visuali... . #rstats | #DataFam | #dataviz | #ggplot2

Two-panel chart titled "Central Manhattan Recorded the Highest Share of Stationary Squirrels." Top panel: stacked bars show behavioral composition by NYC census area from the 2020 Squirrel Census multi-park count — Central Manhattan squirrels were recorded stationary (foraging or eating) 54% of the time, versus 33–34% in Lower Manhattan, Brooklyn, and Upper Manhattan. Bottom panel: a dot plot ranks all 20 sampled parks by stationary share. Four Central Manhattan parks cluster near the top, but St. Nicholas Park (Upper Manhattan) sits within that high-stationary cluster, while John V. Lindsay East River Park (Central Manhattan) falls below the citywide median — showing that the area-level pattern was neither universal within Central Manhattan nor exclusive to it.

Steven Ponce@sponce1· 17d

📊 #TidyTuesday – 2026 W30 | Ecotourism . 🔗: stevenponce.netlify.app/data_visuali... . #rstats | #r4ds | #dataviz | #ggplot2

Paired ridgeline chart comparing observed temperatures with temperature departures from each weather station's seasonal normal for orchids, manta rays, Gouldian finches, and glowworms. The species occupy markedly different raw temperature ranges, but their local temperature anomalies overlap much more. Glowworm is shown as five individual points because only five station-days were available.

Steven Ponce@sponce1· 18d

A routine FDA drug review is now faster than an expedited one was in the 1980s. . Compared review times by pathway and decade: the regular Standard track has overtaken where Priority used to be. . 🔗: stevenponce.netlify.app/projects/sta... . #rstats | #r4ds | #dataviz | #FDA

Editorial infographic titled "What was once exceptional is now ordinary." A 1980s FDA Priority review took a median of 788 days; a 2020s FDA Standard review takes 365 days — less than half the time. A line chart below shows median review days by decade, 1985–2025, for both pathways: Priority stayed faster throughout, but its advantage over Standard widened through the 1990s–2000s before narrowing sharply after 2010. Based on 1,387 NME/BLA approvals. Source: FDA CDER.

Steven Ponce@sponce1· 20d

📊 #MakeoverMonday – 2026 W29 | US Data Center Locations . 🔗: stevenponce.netlify.app/data_visuali... . #rstats | #DataFam | #dataviz | #ggplot2

Progress-bar chart showing the share of US data centers currently operating by size class: 79% of small facilities (0–10 MW) are operating, falling to 51% for medium, 38% for large, 14% for hyperscale, and just 5% for mega campuses (>1,000 MW).

Steven Ponce@sponce1· 21d

📊 #TidyTuesday – 2026 W29 | Near-Death Experiences (NDERF) . 🔗: stevenponce.netlify.app/data_visuali... . #rstats | #r4ds | #dataviz | #ggplot2

A delta bar chart titled "Out-of-body experiences are linked to ESP—not to unity." Two horizontal bars show the percentage-point difference in reporting rates between near-death accounts with and without a reported out-of-body experience. The ESP bar extends 18.4 percentage points to the right, from 5% reporting ESP without an out-of-body experience to 23% with one. The Unity bar extends only 2.7 percentage points to the left, from 18% to 15%, shown in muted beige to signal no meaningful association. A caption notes the ESP link remained after adjusting for experience depth, narrative length, reported clinical death, and narrative language, while no comparable link was found for unity. Data from the NDERF near-death experience archive.

Steven Ponce@sponce1· 29d

📊 #MakeoverMonday – 2026 W28 | Historical Tropical Cyclones . 🔗: stevenponce.netlify.app/data_visuali... . #rstats | #DataFam | #dataviz | #ggplot2

Dumbbell dot plot titled "Traditional Park Amenities Are More Common in Cooler-Climate Cities." For five traditional park amenities — basketball hoops, combined fields, playgrounds, baseball and softball diamonds, and tennis courts — cooler-climate U.S. cities (Cold and Mixed-Humid zones) show higher median per-capita counts than warmer-climate cities (Hot-Humid and Hot-Dry zones), with no exceptions. Ratios range from 1.3 to 1.8 times higher in cooler climates; basketball hoops show the largest raw difference, about 2.2 more hoops per 10,000 residents. Filled circles show the average of the two climate-zone medians within each group; smaller open circles show the individual underlying climate-zone medians. Data from Trust for Public Land's City Park Facts.

Steven Ponce@sponce1· 29d

📊 #TidyTuesday – 2026 W28 | Many penguins . 🔗: stevenponce.netlify.app/data_visuali... . #rstats | #r4ds | #dataviz | #ggplot2

Two-panel scatterplot titled "Bigger Penguins Do Not Simply Have Deeper Bills," using AVONET morphometric data for 76 penguins across six genera. The left panel plots wing length against bill shape ratio (depth divided by length): wing length shows only a weak pooled correlation with bill shape (r = -0.32), and a genus-adjusted regression model finds no significant within-genus relationship (p = 0.39) — genus explains most of the variation instead. Eudyptes (n = 31), highlighted in burgundy, has the deepest bills relative to length despite a mid-range body size, while Aptenodytes (n = 8), the largest genus by wing length, has the shallowest. The right panel shows that tarsus length scales with wing length (r = 0.68), confirming that some traits follow expected body-size scaling, whereas bill shape does not. Data: AVONET (Tobias et al. 2022), via TidyTuesday.

Steven Ponce@sponce1· Jul 6

📊 #MakeoverMonday – 2026 W27 | Historical Tropical Cyclones . 🔗: stevenponce.netlify.app/data_visuali... . #rstats | #DataFam | #dataviz | #ggplot2

Two-panel dot-and-whisker chart titled "Atlantic Hurricanes: The Late-Century Lull." The top panel compares the share of Atlantic storms reaching Category 3+ across three tested periods: the early record (1910–1959) at 21%, the late-century lull (1965–1989) at 10%, and the modern record (1995–2025) at 21%, each shown with 95% confidence intervals and a dashed reference line near 21%. The lull, highlighted in red, is the only period whose interval falls clearly below the reference rate. The bottom panel shows the same rate estimated for all 18 decades from the 1850s to the 2020s in muted gray, with the 1970s and 1980s highlighted in red to provide historical context for the top panel's finding. Data from the US National Hurricane Center via IBTrACS (NOAA NCEI).

Steven Ponce@sponce1· Jul 6

📊 #TidyTuesday – 2026 W27 | UFC Athletes and Fight Data . 🔗: stevenponce.netlify.app/data_visuali... . #rstats | #r4ds | #dataviz | #ggplot2

Two-panel chart titled "Reach, Height, and Age Barely Predict a UFC Winner." Left panel shows win rates for the fighter with a physical advantage: reach (48.1%), height (48.0%), and youth (50.1%), all clustered near a dashed 50% coin-flip line with confidence intervals. Right panel plots betting-market implied probability against actual win rate across five buckets from heavy underdog to heavy favorite, with points falling almost exactly on the diagonal reference line and an average calibration error of about 1 percentage point. Data from the fightr R package (UFC athlete profiles, UFCStats, Kaggle, Octagon API).

Steven Ponce@sponce1· Jul 4

❄️ Powerball's Memory Problem . 68 of 69 numbers have been colder than they are today. . Traced each number's drought history since 2015 to see how today's "coldest" number compares to its own worst stretch. . 🔗: stevenponce.netlify.app/projects/sta... . #rstats | #r4ds | #dataviz | #Powerball

Small-multiples chart of all 69 Powerball numbers, each showing its history of droughts since 2015 as a gray timeline, with a burgundy dot marking its worst drought and a hollow circle marking today. 68 of the 69 numbers have gone longer without hitting at some point in their history than today's longest current drought (48 draws). Three panels are highlighted: No. 23 (today's longest drought), No. 35 (the longest drought on record, 112 draws), and No. 69 (currently setting a new personal record at 44 draws, past its old record of 43).

Steven Ponce@sponce1· Jul 4

27 49 50 51 52 Would these numbers look random to you? Real Powerball drawing, Feb 2019 — 1 in 4 drawings has a pair like this. Tested 3 "suspicious" patterns against pure chance. The lottery isn't the subject. It's the evidence. randomness-docket.netlify.app

Steven Ponce@sponce1· Jul 1

📊 #SWDchallenge July 2026 | have a ball: visualize the World Cup . Hosting is supposed to be an edge. Yet no World Cup host has reached a final since France in 1998. . 🔗 tinyurl.com/jru8jnj9 . #dataviz | #rstats | #DataStorytelling

A two-row tile chart, "Hosting the World Cup Doesn't Take Teams as Far as It Once Did," showed how far each World Cup host advanced at its own tournament, 1930–2022. Twenty-two tiles, one per tournament, ran in chronological order, colored by the host's furthest stage: dark red for finalist or champion, rose for semifinal, light pink for quarterfinal/round of 16, pale gray for group stage. No host reached the final after France won on home soil in 1998, marked by a small gap following that tile. Before 1998, hosts reached the final nearly half the time and always advanced past the group stage; the eleven tournaments since have produced only mid-tier results, including two group-stage eliminations (South Africa 2010, Qatar 2022). The 2002 tile, co-hosted by Japan and South Korea, reflected South Korea's deeper semifinal run rather than Japan's round-of-16 finish. Source: Fjelstul World Cup Database (Joshua C. Fjelstul, Ph.D.).

Steven Ponce@sponce1· Jun 30

📊 #MakeoverMonday – 2026 W26 | The World's Space Launch Sites . 🔗: stevenponce.netlify.app/data_visuali... . #rstats | #DataFam | #dataviz | #ggplot2

Two-panel chart titled "The U.S. Launch Lead Is a Starlink Story." Left panel: grouped bar chart of annual orbital launches, 2016–2026 YTD, comparing China (burgundy) to U.S. launches excluding Starlink (blue), with gray dots showing U.S. totals including Starlink. China's bars exceed the U.S. non-Starlink bars every year from 2018 onward; the gray dots rise sharply after 2022, showing Starlink's growing share of the U.S. total. Right panel: cumulative line chart showing China reaching 544 total launches by 2026 versus 382 for the U.S., excluding Starlink, a gap of 162 missions, while total U.S. launches, including Starlink, reach 784. Annotation notes Starlink alone (402 launches) exceeds the U.S.'s 240-launch lead over China in total figures.

Steven Ponce@sponce1· Jun 29

📊 #TidyTuesday – 2026 W26 | Wreck Inventory of Ireland . 🔗: stevenponce.netlify.app/data_visuali... . #rstats | #r4ds | #dataviz | #ggplot2

Line chart titled "Ireland's Shipwrecks Chronicle the End of the Age of Sail." The x-axis spans 1800 to 1940 by decade; the y-axis shows the share of classified wrecks from 0% to 100%. A blue line representing sail-powered vessels starts near 97% in 1800 and declines gradually, then steeply, crossing below 50% during the 1910s before reaching 4% by the 1940s. A burgundy line representing steam and motor vessels begins near 3% and rises slowly through the 1880s, accelerating sharply after 1900 to reach 96% by the 1940s. The two lines cross between 1910 and 1920, marking the decade when steam became the dominant vessel type in Irish waters. Among records with known vessel type, approximately 6,500 of 17,981 total; unknown classification excluded. Source: Wreck Inventory of Ireland, National Monuments Service.

Steven Ponce@sponce1· Jun 23

📊 #MakeoverMonday – 2026 W25 | UFO Sightings . 🔗: stevenponce.netlify.app/data_visuali... . #rstats | #DataFam | #dataviz | #ggplot2

Horizontal dot plot titled "Darkness 'changes' how UFOs are described." Three words witnesses use for round UFOs are placed by the share of each word's reports that occurred in daylight: "orb" at 19%, "circle" at 29%, and "sphere" at 43%, against a dashed reference line marking the 28% overall daylight rate for all UFO reports. The words form an ordered gradient — luminous descriptions ("orb") skew nocturnal, solid-form descriptions ("sphere") skew toward daylight — showing that how a round UFO is described tracks how much light there was to see it. Based on 305 round-shaped NUFORC reports (orb 57, circle 180, sphere 68).

Steven Ponce@sponce1· Jun 22

📊 #TidyTuesday – 2026 W25 | Papal Encyclicals: Industrial Revolution vs. AI Revolution . 🔗: stevenponce.netlify.app/data_visuali... . #rstats | #r4ds | #dataviz | #ggplot2

Diverging bar chart titled "From what workers are owed to what humans are," comparing the distinctive vocabulary of two papal encyclicals. Words extending left are characteristic of Rerum Novarum (1891, Leo XIII); words extending right are characteristic of Magnifica Humanitas (2026, Leo XIV); bar length is each word's weighted log-odds distinctiveness. The 1891 encyclical's signature words are labor, law, property, private, associations, classes, wages, employer, and workmen. The 2026 encyclical's are dignity, person, digital, AI, technology, data, intelligence, artificial, algorithms, and machines. A note observes that both encyclicals lean heavily on a single word — human — and that what differs between them is the threat to it. Source: Vatican.va (Holy See).

Steven Ponce@sponce1· Jun 16

📊 #MakeoverMonday – 2026 W24 | Global Top Foreign Investments . 🔗: stevenponce.netlify.app/data_visuali... . #rstats | #DataFam | #dataviz | #ggplot2

Horizontal bar chart titled "The World's Largest "Investor Countries" Are Often Financial Hubs," ranking the top 20 countries by foreign direct investment outflows in 2024 (billions of US dollars). Five of the ten largest sources are recognized offshore financial centers, shown in burgundy: Luxembourg ($109B, 4th), Hong Kong ($87B), the British Virgin Islands ($59B), Singapore ($55B), and the Netherlands ($55B). Productive economies are gray, led by the United States ($266B), Japan ($204B), and China ($163B). Luxembourg and the British Virgin Islands each report larger outflows than Germany or India, and the top six countries account for 54% of global FDI outflows. The offshore centers are scattered throughout the ranking rather than clustered at the top. Source: Visual Capitalist via data.world; OFC classification from Garcia-Bernardo et al. (2017).

Steven Ponce@sponce1· Jun 14

📊 #TidyTuesday – 2026 W24 | UK Baby Names . 🔗: stevenponce.netlify.app/data_visuali... . #rstats | #r4ds | #dataviz | #ggplot2

Line chart titled "Half of Scotland's baby boys once shared just 18 names." The vertical axis shows the number of distinct names needed to account for half of all babies born each year from 1974 to 2025, with separate lines for boys and girls, and the gap between them shaded. In 1974, just 18 names accounted for half of baby boys, compared with 35 for girls, making girls' names about 1.9 times as varied. Boys' name diversity then climbed steeply to 77 by 2025 while girls' rose more gently to 85; the two lines reached parity in 2017 at 65 names each before the gap reopened slightly. Both sexes became far more varied overall, but the long-standing gap in name diversity between boys and girls nearly closed. Source: National Records of Scotland.

Steven Ponce@sponce1· Jun 8

📊 #MakeoverMonday – 2026 W23 | Coachella Lineup . 🔗: stevenponce.netlify.app/data_visuali... . #rstats | #DataFam | #dataviz | #ggplot2

Slope chart of the number of artists on Coachella's initial lineup announcement, by day, comparing 2018 and 2026. Friday stays nearly flat, slipping from 55 to 51, a loss of 4. Saturday drops from 57 to 44 and Sunday from 57 to 43, a combined weekend loss of 27. The two weekend lines, shown in rust, plunge steeply while Friday, in gray, holds close to level — the contraction concentrated on Saturday and Sunday.

Steven Ponce@sponce1· Jun 6

📊 #TidyTuesday – 2026 W23 | Films Based on Video Games . 🔗: stevenponce.netlify.app/data_visuali... . #rstats | #r4ds | #dataviz | #ggplot2

Boxplot of Rotten Tomatoes critic scores for 73 video game film adaptations, grouped by release decade from the 1990s to the 2020s. Scores climb steadily: median scores are 17 (1990s), 18 (2000s), 37.5 (2010s), and 51 (2020s), and the entire distribution shifts upward, not just the top films. A horizontal line marks the "Fresh" threshold of 60. The share of films scoring Fresh rises from 11% in the 1990s and just 4% in the 2000s to 27% in the 2010s and 37% in the 2020s, with the sharpest jump between the 2000s and 2010s. The typical adaptation is still rated below Fresh, but good video game movies have gone from rare to fairly common.

Steven Ponce@sponce1· Jun 2

📊 #MakeoverMonday – 2026 W22 | Arsenal 25/26 Champions . 🔗: stevenponce.netlify.app/data_visuali... . #rstats | #DataFam | #dataviz | #ggplot2

Two-panel data visualization titled "Arsenal didn't win by scoring the most goals — they won by conceding the fewest." The left panel is a scatter plot of all 20 Premier League clubs in the 2025–26 season, plotting goals scored (vertical) against goals conceded (horizontal axis reversed, so fewer conceded sit to the right). Arsenal, highlighted in red, sits alone in the top-right corner: 71 goals scored and a league-low 27 conceded. Manchester City scored the most goals (77) but conceded more (35) and finished second. The right panel is a bar chart of Arsenal's points by goals conceded per match: 53 points came in clean-sheet games, 27 when conceding one, and just 5 when conceding two or more — 62% of their title-winning points arrived without conceding. Source: FBref, Premier League 2025–26.

Steven Ponce@sponce1· Jun 1

📊 #SWDchallenge June 2026 | when normal is noteworthy . Air travel feels chaotic. Yet U.S. on-time arrivals trace nearly the same seasonal path every year . 🔗 stevenponce.netlify.app/data_visuali... . #dataviz | #rstats | #DataStorytelling

U.S. airline on-time performance follows a nearly identical seasonal rhythm each year, so the summer dip is expected rather than a warning sign. This line chart shows the typical year — the 2013–2019 monthly average — in burgundy: on-time arrivals sit near 80% in winter, dip to about 76% in June, climb to a peak near 86% in September, then fall back through December. A shaded band marks the full range across all seven years; the typical line stays inside it every month, and the band is narrowest in summer — meaning the worst month is also the most consistent. The y-axis runs from 75% to 90% on-time; the x-axis spans January through December.

Steven Ponce@sponce1· May 30

📊 #TidyTuesday – 2026 W22 | European Parenting Leave Policies . 🔗: stevenponce.netlify.app/data_visuali... . #rstats | #r4ds | #dataviz | #ggplot2

A two-panel chart titled "Europe built parental leave for mothers first," covering 21 European countries from 1970 to 2024. The top lollipop panel shows each country's first year of co-parent leave: maternity leave was already universal in 1970, but co-parent leave came far later — Spain already had it in 1970, most countries adopted only after 2000 (median 2002), and Germany had none through 2024. The bottom step panel shows the cumulative number of countries offering co-parent leave rising to 20 of 21, with a note that 19 of 21 had already adopted before the 2019 EU work-life balance directive. Together, the panels reveal that Europe established maternity leave first and extended leave to co-parents decades later. Source: European Parenting Leave Policies (EPLP) Dataset, Spitzer et al. (2025).

Steven Ponce@sponce1· May 25

📊 #MakeoverMonday – 2026 W21 | GPU computational performance per dollar . 🔗: stevenponce.netlify.app/data_visuali... . #rstats | #DataFam | #dataviz | #ggplot2

A log-scale scatter plot titled "A Dollar Buys Far More AI Computation Than a Decade Ago" showing GPU computational performance per dollar (FLOP/s per dollar, constant 2024 US$) from 2008 to 2023. A navy frontier line connects the best available price-performance GPU at each point in time, rising from roughly 500 million FLOP/s per dollar for the NVIDIA GeForce GTX 280 in 2008 to over 50 billion for the AMD Radeon RX 7900 XTX in 2022 — a roughly 100-fold increase. The steepest acceleration occurs after 2017. Four parchment-colored vertical bands mark AI milestones: AlexNet (2012), Transformer (2017), GPT-2 (2019), and ChatGPT (2022). A muted gray cloud of all 51 GPUs provides context. Each horizontal band represents 10× more compute per dollar. Data from Epoch AI (2026) and U.S. Bureau of Labor Statistics (2026).

Steven Ponce@sponce1· May 23

📊 #TidyTuesday – 2026 W21 | Sustainable Energy for All . 🔗: stevenponce.netlify.app/data_visuali... . #rstats | #r4ds | #dataviz | #ggplot2

Two-panel dumbbell chart titled "Getting Electricity Is Just the First Step." Panel A, "Rural Households Are Still Waiting for the Grid," shows 12 countries where urban electricity access far exceeds rural access — gaps range from roughly 60 to 80 percentage points, with Guinea, Ethiopia, and Angola showing the largest divides. Panel B, "Having Electricity Doesn't Mean Cooking Cleanly," shows 12 countries with at least 85% electricity access, where clean cooking fuel access lags far behind — Sri Lanka, Mongolia, and Pakistan lead with gaps of nearly 60 percentage points. Both panels use burgundy segments to visualize the gap between a lighter dot (the lagging metric) and a darker dot (the achieved metric). The subtitle notes that millions still cook with wood, charcoal, or dung — a major source of indoor air pollution and disease.

Steven Ponce@sponce1· May 19

📊 #MakeoverMonday – 2026 W20 | UK Fuel Prices . 🔗: stevenponce.netlify.app/data_visuali... . #rstats | #DataFam | #dataviz | #ggplot2

A two-panel line chart titled "Diesel's Crisis Premium." The top panel shows indexed UK fuel prices (Jan 2013 = 100) for diesel and petrol from 2013 to 2026, with both fuels moving largely in tandem before diverging sharply during crisis periods. The bottom panel isolates the diesel premium over petrol in pence per liter, revealing that during stable years, the gap stayed near the pre-2022 average of roughly +3.7p. The Ukraine conflict in 2022 pushed the premium to a peak of +24.5p; the 2026 Iran conflict drove it higher still, reaching +31.9p — the widest recorded spread in the dataset. Soft gray shading marks both crisis windows across panels. Data: RAC Fuel Watch via MakeoverMonday 2026 W20.

Steven Ponce@sponce1· May 18

📊 #TidyTuesday – 2026 W20 | Crossref Metadata Coverage . 🔗: stevenponce.netlify.app/data_visuali... . #rstats | #r4ds | #dataviz | #ggplot2