Talk:COVID-19 pandemic in Sweden/Archive 4

This is an archive of past discussions about COVID-19 pandemic in Sweden. Do not edit the contents of this page. If you wish to start a new discussion or revive an old one, please do so on the current talk page.

Archive 1

Archive 2

Archive 3

Archive 4

Archive 5

Excess mortality - per region distribution

Latest comment: 3 years ago6 comments2 people in discussion

Are there any concerns against adding weekly deaths per region charts (with or without smoothening)?

All data was taken from official source: [1], [2]

In below charts regions are sorted basically by death numbers range, and secondary, within range - alphabetically. When couples of regions was plotted on the same chart - it was made basically by "random combinations" - just to achieve better visual readability. But first 3-pl (Stockholm, Västra Götaland, Skåne) was shown together because they are the naim "contributors" to total numbers. 84.47.179.91 (talk) 10:30, 19 August 2020 (UTC)

All-cause weekly deaths 2015–2021 per regions (without smoothening)

All-cause weekly deaths 2015–2021 per regions (smoothed by 4-weeks rolling average)

All-cause daily deaths 2015–2021 weekly per regions (source code of script that was used for markup rendering)

Source code of script for converting local XLSX file with statistics into wiki charts mockup (js+html)

<html><head></head>
<body>

<div>
  download and apply
  <a href="https://www.socialstyrelsen.se/globalassets/1-globalt/covid-19-statistik/statistik-relaterad-till-covid-19/antal-doda--per-lan-och-vecka-2015-v1-2020.xlsx">
  antal-doda--per-lan-och-vecka-2015-v1-2020.xlsx
  </a>
  (<a href="https://www.socialstyrelsen.se/statistik-och-data/statistik/statistik-om-covid-19/statistik-relaterad-till-covid-19/">
  [1]
  </a>)
</div>

<input type="file" id="xlsxFileIn">
<button id="xlsxFilePraseBt">parse</button>
addDateUnknown:<input type="checkbox" id="addDateUnknownCb">
use4WeeksRollAvg:<input type="checkbox" id="use4wRollAvgCb"><br/>
<textarea id="chartsMdOut" style="width:90%;height:150px"></textarea>

<script src="https://ajax.googleapis.com/ajax/libs/jquery/2.1.1/jquery.min.js"></script>
<script src="https://cdnjs.cloudflare.com/ajax/libs/xlsx/0.8.0/jszip.js"></script>
<script src="https://cdnjs.cloudflare.com/ajax/libs/xlsx/0.8.0/xlsx.js"></script>
<script>

  xlsxFilePraseBt.onclick = async () => {
    var files = xlsxFileIn.files;
    if (!files.length) return;
    var cfg = {
      addDateUnknown: addDateUnknownCb.checked, use4wRollAvg: use4wRollAvgCb.checked,
      chartYAxisTitle: "Deaths per week"
    };
    chartsMdOut.value = (await parseExcelAndRenderMd(files[0], cfg)).join("\n");
  }

  var parseExcelAndRenderMd = async (file, cfg) => {
    var workbook = window.workbook = await readExcelWb(file);
    workbook.SheetNames.forEach((sheetName) => {
      console.log(sheetName);
    });
    var regionsAllData = parseStatistikenPerLanWb(workbook);
    return formatChartsMd(regionsAllData, cfg);
  }


  var assert = console.assert;

  var readExcelWb = async (file) => (
    XLSX.read(
      new Uint8Array(await file.arrayBuffer()), 
      {type: 'array'}
    )
  );

  var cellText = (cell) => ((cell && cell.w) || XLSX.utils.format_cell(cell))
  var cellNum = (cell) => {
    assert(cell && cell.t == "n", cell);
    return (parseFloat((cell && cell.v) || 0));
  };

  var parseStatistikenPerLanWb = (wb) => {
    assert(wb.SheetNames.length == 23, wb.SheetNames.length);
    assert(wb.SheetNames[0] == "Om statistiken", wb.SheetNames[0]);
    var regions = wb.SheetNames.slice(1);

    return regions.map((region) => {
      var sheet = wb.Sheets[region];

      assert(cellText(sheet.A5) == "Vecka", sheet.A5);
      assert(cellText(sheet.A59).startsWith("ej fastställt"), sheet.A59);

      var years = [];
      var yearColId = ["B","C","D","E","F","G", "H"];
      for (var c = 0; c <= 6; ++c) {
        var year = 2015 + c, cId = yearColId[c];

        assert(cellText(sheet[cId+"5"]) == year, sheet[cId+"5"], year);
        var weeks = [];
        for (var r = 0; r < 53; ++r) {
          var rId = 6 + r;
          assert(cellNum(sheet["A"+rId]) == r+1, sheet["A"+rId], r+1);
          var cell = sheet[cId + rId];
          if (cell == null) break;
          weeks.push(cellNum(cell));
        }
        assert(year == 2021 || weeks.length >= 52);
        var dateUnknown = cellNum(sheet[cId+59]);
        var errPerWeek = dateUnknown / weeks.length;
        var unknownPerWeek = weeks.map((_,w)=>(
          Math.round(errPerWeek*(w+1)) - Math.round(errPerWeek*(w))
        ));
        years.push({
          id: cellText(sheet[cId+"5"]),
          weeks: weeks,
          dateUnknown: dateUnknown,
          unknownPerWeek: unknownPerWeek
        });
      }
      return {
        name: region,
        years: years
      };
    });
  };

  var ad7d = (date, n) => (n=Number.isFinite(n)?n:1, new Date(date.getTime()+n*7*24*60*60*1000));
  //var weekKey = (w0date, n) => (`w${n+1} ${ad7d(w0date, n).toJSON().slice(0,10)}`);
  var weekKey = (w0date, n) => (ad7d(w0date, n).toJSON().slice(0,10).replace(/-/g,"/"));

  var formatChartMd = (regionsAllDataArr, regionsNames, cfg) => {
    var {addDateUnknown, use4wRollAvg, useStackedArea, chartColors, chartYAxisTitle} = cfg;
    var addErr = !!addDateUnknown;
    var useRollAvg = !!use4wRollAvg;
    var chartType = useStackedArea ? "stackedarea" : "line";
    var rollAvgTitleSufix = "(4-weeks rolling average)";
    var chartYAxisTitle = useRollAvg ? `${chartYAxisTitle} ${rollAvgTitleSufix}` : chartYAxisTitle;

    var regionsAllData = Object.fromEntries(regionsAllDataArr.map((r) => ([r.name, r])));
    var yearW0 = [
      new Date("2014-12-29"), new Date("2016-01-04"), new Date("2017-01-02"), 
      new Date("2018-01-01"), new Date("2018-12-31"), new Date("2019-12-30"), 
      new Date("2021-01-04")
    ];
    var regionsData = regionsNames.map((regName)=>(regionsAllData[regName]));

    var yearsWeeksK = [];
    var yearsWeeksReg = regionsData.map(()=>([]));
    for (var yId = 0; yId <= 6; ++yId) {
      var year = 2015 + yId;
      var regYear = regionsData[0].years[yId];
      assert(regYear.id == year, regYear.id, year);
      var weeksK = regYear.weeks.map((_,i)=>(weekKey(yearW0[yId], i+1)));
      yearsWeeksK.push(weeksK);
    }
    var yearsWeeksReg = regionsData.map((region)=>(
      region.years.flatMap((y)=>(y.weeks.map(
        (v,i) => (!addErr? v : v + y.unknownPerWeek[i])
      )))
    ));
    if (useRollAvg) {
      var rnd4 = (v) => (Math.round(v*10000)/10000);
      var rollAvg = (arr, pos, win) => (
        arr = arr.slice(Math.max(0, pos+1 - win), pos+1),
        rnd4(arr.reduce((a, b) => (a + b), 0) / Math.min(win, Math.max(1, arr.length)))
      );
      yearsWeeksReg = yearsWeeksReg.map((regWeeks) => (
        regWeeks.map((v, i) => (rollAvg(regWeeks,i,4)))
      ));
    }
    var regTitles = regionsData.map((region, regIndex)=>(
      ` | y${regIndex+1}Title=${region.name}`
    ));
    var mdXValues = ` | x=${yearsWeeksK.flat().join()}`;
    var regMdYValues = yearsWeeksReg.map((regWeeks, regIndex)=>(
      ` | y${regIndex+1}=${regWeeks.join()}`
    ));

    var chartMd = 
`
{{Graph:Chart
 | width = 1000
 | height = 250
 | type = ${chartType}
 | xType = date
 | yAxisMin = 0
 | yAxisTitle = ${chartYAxisTitle}
${chartColors &&
` | colors=${chartColors}` || ""
}
 | legend=Regions
${[...regTitles, mdXValues, ...regMdYValues].join("\n")}
}}
`;
    return chartMd;
  };

  var formatChartsMd = (regionsAllDataArr, cfg) => {
    var {addDateUnknown, use4wRollAvg} = cfg;
    console.log("addDateUnknown", !!addDateUnknown, "use4wRollAvg", !!use4wRollAvg);
    var regionsNamesGr0 = [
      ["Stockholm","Västra Götaland","Skåne"],
      ["Gävleborg","Halland","Jönköping","Norrbotten","Uppsala","Värmland",
       "Västernorrland","Örebro","Östergötland"],
      ["Dalarna","Kalmar","Södermanland","Västerbotten","Västmanland"],
      ["Blekinge","Jämtland","Kronoberg"],
      ["Gotland"]
    ];
    var regionsNamesGr = [
      ["Skåne","Stockholm","Västra Götaland"],
      ["Södermanland","Västmanland"],
      ["Dalarna","Jönköping"],
      ["Gävleborg","Norrbotten"],
      ["Uppsala","Värmland"],
      ["Gotland","Jämtland","Västernorrland","Östergötland"],
      ["Blekinge","Västerbotten","Örebro"],
      ["Halland","Kalmar","Kronoberg"]
    ];
    var overallChartMd = formatChartMd(
      regionsAllDataArr, regionsNamesGr0.flat(), {...cfg, useStackedArea: true,
        chartColors: "#ff7f0e,#ffbb78,#1f77b4,#aec7e8,#2ca02c,#98df8a,#d62728,#ff9896,#9467bd,#c5b0d5,#8c564b,#c49c94,#e377c2,#f7b6d2,#7f7f7f,#c7c7c7,#bcbd22,#dbdb8d,#17becf,#9edae5"
      }
    );

    var chartsMd = [overallChartMd, ...regionsNamesGr.map((regionsNames) => (
      formatChartMd(regionsAllDataArr, regionsNames, cfg)
    ))];
    console.log(chartsMd.join("\n"))
    return chartsMd;
  };

</script>

</body></html>

created: 84.47.179.91 (talk) 10:56, 19 August 2020 (UTC)

updated: 84.47.179.91 (talk) 14:24, 5 April 2021 (UTC)

To include all of the above seems like a massive overkill. The idea of charting in some way the differences between regions seems fine. Adding a single chart showing all-cause deaths for top 3 affected regions by death rate seems fine, but the colors should be toned down to be much calmer; especially the orange yells too much. I am thinking of adding the following single chart I picked from the above:

I set the chart width to 700 since 1000 causes the chart to be too wide on some devices.

Are Skåne, Stockholm and Västra Götaland the top 3 by deaths per capita or by absolute deaths? Deaths per capita are more relevant than absolute deaths. (I could find that myself, but I am tired.)

The data added above to the thread has over 160 kb; that's a lot even for the talk page. --Dan Polansky (talk) 10:00, 21 August 2020 (UTC)

Reading colors - I've just used default paletta, only little bit reorder it for first chart. So I think other palate could be good enough.

Answering your question about Skåne, Stockholm and Västra Götaland - it was picked up by absolute values (not by per capita values). However nearby excel file (antal-doda-per-100000-per-lan-och-vecka-2015-v1-2020.xlsx) from same source contains per capita values, and has just the same structure, so the same script could be used to convert it to wiki charts markup. And yes, in general I agree with your statement that "Deaths per capita are more relevant than absolute deaths." but here for me it was interesting to show which regions contributed the most to total values (for entire country). And by arranging regions in stacked chart, I was trying to show that without topmost (Stockholm) layer situation in Sweden this year looks pretty comparable with previous years.

Regarding markup size - it is true it is big enough ... (Maybe charts could be extracted to separate template, like Template:COVID-19_pandemic_data/Sweden_medical_cases. However I'm not sure whether it is good idea or not). Here by plotting all charts I've just tried to demonstrate where Swedish strategy works well, and where it face with some problems. So, looking at all charts I could clearly see irregular peaks for Stockholm, Södermanland, Västmanland, and less irregular but still considerable peak for Dalarna. But honestly, I could not see any irregular peaks for all other regions. However I don't want to make any conclusions instead of readers of this materials. 84.47.179.91 (talk) 11:34, 21 August 2020 (UTC)

@Dan Polansky: as you'd suggested, it was also added graph with per capita values. When plotting many regions together graphes quickly becomes massy, but I still found very interesting following chart which brings more context to per capita values in Stockholm (based on [3])

Of course here Norrbotten county was picked up as example of (Swedish) worst per capita mortality in nearest past. But for other regions numbers in range 25 - 30 was also reachable in the past. (Better understanding of that numbers could bring information about demography in those regions, and/or breakdown of all cases into age groups) 84.47.179.91 (talk) 11:46, 26 August 2020 (UTC)