Vega lite calculate issues

Question

I am trying to create a graph that displays a new column values calculated from many existing columns in my csv file. I know that there are way more countries and that they all have a score out of 10 (for each field) so out of 50 total.

     {
  "$schema": "https://vega.github.io/schema/vega-lite/v4.json",
  "data": {
    "url": "https://raw.githubusercontent.com/Anika6138/InfoVis/master/arabica_data_cleaned.csv",
    "format": {
      "type": "csv"
    }
  },
  "transform": [
   {"calculate": "datum.Aroma + datum.Flavor + datum.Aftertaste + datum.Acidity + datum.Sweetness  ", "as": "Taste_Points"}
  ],
      "mark": "bar",
      "encoding": {
        "y": {
          "field": "Country_of_Origin",
          "type": "nominal"
        },
        "x": {
          "field": "Taste_Points",
          "type": "quantitative"
        }
  },
  "config": {}
}

This is what I get. Many countries with values are ignored and no filters are added.

Answer 1

Your data is specified as CSV, which means all values in calculations are interpreted as strings unless you specify otherwise. There are two ways to fix this; you can add a parse statement in the data format definition:

  "data": {
    "url": "https://raw.githubusercontent.com/Anika6138/InfoVis/master/arabica_data_cleaned.csv",
    "format": {
      "type": "csv",
      "parse": {"Aroma": "number", "Flavor": "number", "Aftertaste": "number", "Acidity": "number", "Sweetness": "number"}
    }
  }

or you can use parseFloat within the calculate expression:

  "transform": [
    {
      "calculate": "parseFloat(datum.Aroma) + parseFloat(datum.Flavor) + parseFloat(datum.Aftertaste) + parseFloat(datum.Acidity) + parseFloat(datum.Sweetness)",
      "as": "Taste_Points"
    }
  ]

The reason fields were implicitly filtered in your original specification is because the result of the sum was, in many cases, a concatenated string that could not be parsed as a valid number, and NaN values are implicitly removed from quantitative encodings; for example:

{
  "data": {
    "values": [
      {"y": "A", "x": 1},
      {"y": "B", "x": 2},
      {"y": "C", "x": null},
      {"y": "D", "x": null}
    ]
  },
  "mark": "bar",
  "encoding": {
    "x": {"field": "x", "type": "quantitative"},
    "y": {"field": "y", "type": "ordinal"}
  }
}