{-# LANGUAGE OverloadedStrings #-} -- -- Based on the Elm VegaLite GalleryAdvanced.elm (from development of version -- 1.13.0) -- module Gallery.Advanced (testSpecs) where import Graphics.Vega.VegaLite import Prelude hiding (filter, lookup) import Data.Aeson (Value, toJSON) testSpecs :: [(String, VegaLite)] testSpecs = [ ("advanced1", advanced1) , ("advanced2", advanced2) , ("advanced3", advanced3) , ("advanced4", advanced4) , ("advanced5", advanced5) , ("advanced6", advanced6) , ("advanced7", advanced7) , ("advanced8", advanced8) , ("advanced9", advanced9) , ("density1", density1) , ("density2", density2) , ("filter1", filter1) , ("filter2", filter2) , ("transform1", transform1) , ("layered1", layered1) , ("layered2", layered2) , ("layered3", layered3) , ("benchmark", benchmark) ] -- helpers -- pQuant, pNominal, pOrdinal, pTemporal :: PositionChannel pQuant = PmType Quantitative pNominal = PmType Nominal pOrdinal = PmType Ordinal pTemporal = PmType Temporal fNominal :: FacetChannel fNominal = FmType Nominal mNominal :: MarkChannel mNominal = MmType Nominal advanced1 :: VegaLite advanced1 = let desc = description "Calculation of percentage of total" dvals = dataFromColumns [] . dataColumn "Activity" (Strings [ "Sleeping", "Eating", "TV", "Work", "Exercise" ]) . dataColumn "Time" (Numbers [ 8, 2, 4, 8, 2 ]) trans = transform . window [ ( [ WAggregateOp Sum, WField "Time" ], "TotalTime" ) ] [ WFrame Nothing Nothing ] . calculateAs "datum.Time/datum.TotalTime * 100" "PercentOfTotal" enc = encoding . position X [ PName "PercentOfTotal", PmType Quantitative, PAxis [ AxTitle "% of total time" ] ] . position Y [ PName "Activity", PmType Nominal ] in toVegaLite [ desc, dvals [], trans [], mark Bar [], heightStep 12, enc [] ] advanced2 :: VegaLite advanced2 = let desc = description "Calculation of difference from average" dvals = dataFromUrl "https://vega.github.io/vega-lite/data/movies.json" trans = transform . filter (FExpr "isValid(datum.IMDB_Rating)") . window [ ( [ WAggregateOp Mean, WField "IMDB_Rating" ], "AverageRating" ) ] [ WFrame Nothing Nothing ] . filter (FExpr "(datum.IMDB_Rating - datum.AverageRating) > 2.5") barEnc = encoding . position X [ PName "IMDB_Rating", PmType Quantitative, PAxis [ AxTitle "IMDB Rating" ] ] . position Y [ PName "Title", PmType Ordinal ] barSpec = asSpec [ mark Bar [], barEnc [] ] ruleEnc = encoding . position X [ PName "AverageRating", PAggregate Mean, PmType Quantitative ] ruleSpec = asSpec [ mark Rule [ MColor "red" ], ruleEnc [] ] in toVegaLite [ desc, dvals [], trans [], layer [ barSpec, ruleSpec ] ] advanced3 :: VegaLite advanced3 = let desc = description "Calculation of difference from annual average" dvals = dataFromUrl "https://vega.github.io/vega-lite/data/movies.json" [ Parse [ ( "Release_Date", FoDate "%d-%b-%y" ) ] ] trans = transform . filter (FExpr "isValid(datum.IMDB_Rating)") . timeUnitAs Year "Release_Date" "year" . window [ ( [ WAggregateOp Mean, WField "IMDB_Rating" ], "AverageYearRating" ) ] [ WGroupBy [ "year" ], WFrame Nothing Nothing ] . filter (FExpr "(datum.IMDB_Rating - datum.AverageYearRating) > 2.5") barEnc = encoding . position X [ PName "IMDB_Rating", PmType Quantitative, PAxis [ AxTitle "IMDB Rating" ] ] . position Y [ PName "Title", PmType Ordinal ] barSpec = asSpec [ mark Bar [ MClip True ], barEnc [] ] tickEnc = encoding . position X [ PName "AverageYearRating", PmType Quantitative ] . position Y [ PName "Title", PmType Ordinal ] . color [ MString "red" ] tickSpec = asSpec [ mark Tick [], tickEnc [] ] in toVegaLite [ desc, dvals, trans [], layer [ barSpec, tickSpec ] ] advanced4 :: VegaLite advanced4 = let desc = description "A scatterplot showing each movie in the database and the difference from the average movie rating." dvals = dataFromUrl "https://vega.github.io/vega-lite/data/movies.json" trans = transform . filter (FExpr "isValid(datum.IMDB_Rating)") . filter (FRange "Release_Date" (DateRange [] [ DTYear 2019 ])) . window [ ( [ WAggregateOp Mean, WField "IMDB_Rating" ], "AverageRating" ) ] [ WFrame Nothing Nothing ] . calculateAs "datum.IMDB_Rating - datum.AverageRating" "RatingDelta" enc = encoding . position X [ PName "Release_Date", PmType Temporal ] . position Y [ PName "RatingDelta" , PmType Quantitative , PAxis [ AxTitle "Residual" ] ] in toVegaLite [ desc , dvals [] , trans [] , enc [] , mark Point [ MStrokeWidth 0.3, MOpacity 0.3 ] ] advanced5 :: VegaLite advanced5 = let des = description "Line chart showing ranks over time for thw World Cup 2018 Group F teams" dvals = dataFromColumns [] . dataColumn "team" (Strings [ "Germany", "Mexico", "South Korea", "Sweden", "Germany", "Mexico", "South Korea", "Sweden", "Germany", "Mexico", "South Korea", "Sweden" ]) . dataColumn "matchday" (Numbers [ 1, 1, 1, 1, 2, 2, 2, 2, 3, 3, 3, 3 ]) . dataColumn "point" (Numbers [ 0, 3, 0, 3, 3, 6, 0, 3, 3, 6, 3, 6 ]) . dataColumn "diff" (Numbers [ -1, 1, -1, 1, 0, 2, -2, 0, -2, -1, 0, 3 ]) trans = transform . window [ ( [ WOp Rank ], "rank" ) ] [ WSort [ WDescending "point", WDescending "diff" ], WGroupBy [ "matchday" ] ] enc = encoding . position X [ PName "matchday", PmType Ordinal ] . position Y [ PName "rank", PmType Ordinal ] . color [ MName "team", MmType Nominal, MScale teamColours ] teamColours = categoricalDomainMap [ ( "Germany", "black" ) , ( "Mexico", "#127153" ) , ( "South Korea", "#c91a3c" ) , ( "Sweden", "#0c71ab" ) ] in toVegaLite [ des , title "World Cup 2018: Group F Rankings" [ TFrame FrBounds, TFontStyle "italic" ] , dvals [] , trans [] , enc [] , mark Line [ MOrient Vertical ] ] advanced6 :: VegaLite advanced6 = let des = description "Waterfall chart of monthly profit and loss" dvals = dataFromColumns [] . dataColumn "label" (Strings [ "Begin", "Jan", "Feb", "Mar", "Apr", "May", "Jun", "Jul", "Aug", "Sep", "Oct", "Nov", "Dec", "End" ]) . dataColumn "amount" (Numbers [ 4000, 1707, -1425, -1030, 1812, -1067, -1481, 1228, 1176, 1146, 1205, -1388, 1492, 0 ]) trans = transform . window [ ( [ WAggregateOp Sum, WField "amount" ], "sum" ) ] [] . window [ ( [ WOp Lead, WField "label" ], "lead" ) ] [] . calculateAs "datum.lead === null ? datum.label : datum.lead" "lead" . calculateAs "datum.label === 'End' ? 0 : datum.sum - datum.amount" "previous_sum" . calculateAs "datum.label === 'End' ? datum.sum : datum.amount" "amount" . calculateAs "(datum.label !== 'Begin' && datum.label !== 'End' && datum.amount > 0 ? '+' : '') + datum.amount" "text_amount" . calculateAs "(datum.sum + datum.previous_sum) / 2" "center" . calculateAs "datum.sum < datum.previous_sum ? datum.sum : ''" "sum_dec" . calculateAs "datum.sum > datum.previous_sum ? datum.sum : ''" "sum_inc" enc = encoding . position X [ PName "label", PmType Ordinal, PSort [], PTitle "Months" ] enc1 = encoding . position Y [ PName "previous_sum", PmType Quantitative, PTitle "Amount" ] . position Y2 [ PName "sum" ] . color [ MDataCondition [ ( Expr "datum.label === 'Begin' || datum.label === 'End'", [ MString "#f7e0b6" ] ) , ( Expr "datum.sum < datum.previous_sum", [ MString "#f78a64" ] ) ] [ MString "#93c4aa" ] ] spec1 = asSpec [ enc1 [], mark Bar [ MSize 45 ] ] enc2 = encoding . position X2 [ PName "lead" ] . position Y [ PName "sum", PmType Quantitative ] spec2 = asSpec [ enc2 [] , mark Rule [ MColor "#404040" , MOpacity 1 , MStrokeWidth 2 , MXOffset (-22.5) , MX2Offset 22.5 ] ] enc3 = encoding . position Y [ PName "sum_inc", PmType Quantitative ] . text [ TName "sum_inc", TmType Nominal ] spec3 = asSpec [ enc3 [] , mark Text [ MdY (-8) , MFontWeight Bold , MColor "#404040" ] ] enc4 = encoding . position Y [ PName "sum_dec", PmType Quantitative ] . text [ TName "sum_dec", TmType Nominal ] spec4 = asSpec [ enc4 [] , mark Text [ MdY 8 , MBaseline AlignTop , MFontWeight Bold , MColor "#404040" ] ] enc5 = encoding . position Y [ PName "center", PmType Quantitative ] . text [ TName "text_amount", TmType Nominal ] . color [ MDataCondition [ ( Expr "datum.label === 'Begin' || datum.label === 'End'" , [ MString "#725a30" ] ) ] [ MString "white" ] ] spec5 = asSpec [ enc5 [], mark Text [ MBaseline AlignMiddle, MFontWeight Bold ] ] in toVegaLite [ des , width 800 , height 450 , dvals [] , trans [] , enc [] , layer [ spec1, spec2, spec3, spec4, spec5 ] ] -- advanced9 in elm advanced7 :: VegaLite advanced7 = let des = description "Using the lookup transform to combine data" groupData = dataFromUrl "https://vega.github.io/vega-lite/data/lookup_groups.csv" [] peopleData = dataFromUrl "https://vega.github.io/vega-lite/data/lookup_people.csv" [] trans = transform . lookup "person" peopleData "name" (LuFields [ "age", "height" ]) enc = encoding . position X [ PName "group", PmType Ordinal ] . position Y [ PName "age", PmType Quantitative, PAggregate Mean ] in toVegaLite [ des, groupData, trans [], enc [], mark Bar [] ] -- advanced14 in elm advanced8 :: VegaLite advanced8 = let des = description "Parallel coordinates plot with manual generation of parallel axes" cfg = configure . configuration (ViewStyle [ ViewNoStroke ]) . configuration (AxisX [ Domain False, LabelAngle 0, TickColor "#ccc" ]) . configuration (MarkNamedStyles [ ( "label", [ MBaseline AlignMiddle, MAlign AlignRight, MdX (-5), MTooltip TTNone ] ) , ( "tick", [ MOrient Horizontal, MTooltip TTNone ] ) ] ) dvals = dataFromUrl "https://vega.github.io/vega-lite/data/iris.json" fields = ["petalLength", "petalWidth", "sepalLength", "sepalWidth"] trans = transform . window [ ( [ WAggregateOp Count ], "index" ) ] [] . fold fields . joinAggregate [ opAs Min "value" "min", opAs Max "value" "max" ] [ WGroupBy [ "key" ] ] . calculateAs "(datum.value - datum.min) / (datum.max-datum.min)" "normVal" . calculateAs "(datum.min + datum.max) / 2" "mid" encLine = encoding . position X [ PName "key", PmType Nominal ] . position Y [ PName "normVal", PmType Quantitative, PAxis [] ] . color [ MName "species", MmType Nominal ] . detail [ DName "index", DmType Nominal ] . tooltips (map (\n -> [TName n, TmType Quantitative]) fields) specLine = asSpec [ encLine [], mark Line [ MOpacity 0.3 ] ] encAxis = encoding . position X [ PName "key", PmType Nominal, PAxis [ AxNoTitle ] ] . detail [ DAggregate Count, DmType Quantitative ] specAxis = asSpec [ encAxis [], mark Rule [ MColor "#ccc" ] ] encAxisLabelsTop = encoding . position X [ PName "key", PmType Nominal ] . position Y [ PNumber 0 ] . text [ TName "max", TmType Quantitative, TAggregate Max ] specAxisLabelsTop = asSpec [ encAxisLabelsTop [], mark Text [ MStyle [ "label" ] ] ] encAxisLabelsMid = encoding . position X [ PName "key", PmType Nominal ] . position Y [ PNumber 150 ] . text [ TName "mid", TmType Quantitative, TAggregate Min ] specAxisLabelsMid = asSpec [ encAxisLabelsMid [], mark Text [ MStyle [ "label" ] ] ] encAxisLabelsBot = encoding . position X [ PName "key", PmType Nominal ] . position Y [ PHeight ] . text [ TName "min", TmType Quantitative, TAggregate Min ] specAxisLabelsBot = asSpec [ encAxisLabelsBot [], mark Text [ MStyle [ "label" ] ] ] in toVegaLite [ des , cfg [] , width 600 , height 300 , dvals [] , trans [] , layer [ specLine, specAxis, specAxisLabelsTop, specAxisLabelsMid, specAxisLabelsBot ] ] -- advanced15 in elm advanced9 :: VegaLite advanced9 = let desc = description "Production budget of the film with highest US Gross in each major genre." dvals = dataFromUrl "https://vega.github.io/vega-lite/data/movies.json" enc = encoding . position X [ PName "Production_Budget" , PmType Quantitative , PAggregate (ArgMax (Just "US_Gross")) ] . position Y [ PName "Major_Genre", PmType Nominal ] in toVegaLite [ desc, dvals [], enc [], mark Bar [] ] -- advanced12 in elm -- -- changed slightly to match naming used by -- https://vega.github.io/vega-lite/docs/density.html#example-faceted-density-estimates -- density1 :: VegaLite density1 = let dvals = dataFromUrl "https://vega.github.io/vega-lite/data/iris.json" [] trans = transform . foldAs [ "petalWidth", "petalLength", "sepalWidth", "sepalLength" ] "organ" "value" . density "value" [ DnBandwidth 0.3, DnGroupBy [ "organ" ] ] enc = encoding . position X [ PName "value", pQuant, PTitle "value (cm)" ] . position Y [ PName "density", pQuant ] . row [ FName "organ", fNominal ] in toVegaLite [ width 300, height 50, dvals, trans [], enc [], mark Area [] ] -- advanced13 in elm -- density2 :: VegaLite density2 = let dvals = dataFromUrl "https://vega.github.io/vega-lite/data/iris.json" [] trans = transform . foldAs [ "petalWidth", "petalLength", "sepalWidth", "sepalLength" ] "measurement" "value" . density "value" [ DnBandwidth 0.3 , DnGroupBy [ "measurement" ] , DnExtent 0 8 , DnSteps 200 ] enc = encoding . position X [ PName "value", pQuant, PTitle "width/length (cm)" ] . position Y [ PName "density", pQuant ] . color [ MName "measurement", mNominal ] in toVegaLite [ width 400, height 100, dvals, trans [], enc [], mark Area [ MOpacity 0.5 ] ] -- advanced7 in elm filter1 :: VegaLite filter1 = let des = description "Filtering the top-k items" dvals = dataFromColumns [] . dataColumn "student" (Strings [ "A", "B", "C", "D", "E", "F", "G", "H", "I", "J", "K", "L", "M", "N", "O", "P", "Q", "R", "S", "T", "U", "V" ]) . dataColumn "score" (Numbers [ 100, 56, 88, 65, 45, 23, 66, 67, 13, 12, 50, 78, 66, 30, 97, 75, 24, 42, 76, 78, 21, 46 ]) trans = transform . window [ ( [ WOp Rank ], "rank" ) ] [ WSort [ WDescending "score" ] ] . filter (FExpr "datum.rank <=5") enc = encoding . position X [ PName "score", pQuant ] . position Y [ PName "student" , pNominal -- , PSort [ ByFieldOp "score" Mean, Descending ] , PSort [ ByChannel ChX, Descending ] ] in toVegaLite [ des, dvals [], trans [], enc [], mark Bar [] ] -- advanced8 in elm filter2 :: VegaLite filter2 = let des = description "Top-k items with 'others'" dvals = dataFromUrl "https://vega.github.io/vega-lite/data/movies.json" [] trans = transform . aggregate [ opAs Mean "Worldwide_Gross" "aggregateGross" ] [ "Director" ] . window [ ( [ WOp RowNumber ], "rank" ) ] [ WSort [ WDescending "aggregateGross" ] ] . calculateAs "datum.rank < 10 ? datum.Director : 'All Others'" "rankedDirector" enc = encoding . position X [ PName "aggregateGross" , pQuant , PAggregate Mean , PNoTitle ] . position Y [ PName "rankedDirector" , pOrdinal , PSort [ ByChannel ChX, Descending ] , PNoTitle ] in toVegaLite [ des , title "Top Directors by Average Worldwide Gross" [] , dvals , trans [] , enc [] , mark Bar [] ] -- advanced10 in elm transform1 :: VegaLite transform1 = let des = description "Cumulative Frequency Distribution" dvals = dataFromUrl "https://gicentre.github.io/data/putneyAirQuality2018.csv" [ Parse [ ( "NOX", FoNumber ) ] ] trans = transform . window [ ( [ WAggregateOp Count ], "cumulativeCount" ) ] [ WSort [ WAscending "NOX" ] ] enc = encoding . position X [ PName "NOX" , pQuant , PTitle "NOX concentration (μg/㎥)" ] . position Y [ PName "cumulativeCount", pQuant ] grStops = [ (0, "white"), (1, "black") ] orangeStops = [ (0, "orange"), (0.5, "green"), (1, "purple") ] -- setting these doesn't really make a difference grOpts = [ GrX1 0, GrX2 1, GrY1 1, GrY2 1 ] markOpts = [ MFillOpacity 0.5 , MFillGradient GrLinear grStops grOpts , MStroke "orange" , MStrokeWidth 2 , MStrokeOpacity 0.8 , MStrokeDash [10, 6, 6, 6] , MStrokeDashOffset 5 , MStrokeGradient GrLinear orangeStops [] ] in toVegaLite [ des, width 500, dvals, trans [], enc [], mark Area markOpts ] -- advanced11 in elm layered1 :: VegaLite layered1 = let des = description "Layered Histogram and Cumulative Histogram" dvals = dataFromUrl "https://vega.github.io/vega-lite/data/movies.json" [] trans = transform . binAs [] "IMDB_Rating" "binIMDB_Rating" . aggregate [ opAs Count "" "count" ] [ "binIMDB_Rating", "binIMDB_Rating_end" ] . filter (FExpr "datum.binIMDB_Rating !== null") . window [ ( [ WAggregateOp Sum, WField "count" ], "cumulativeCount" ) ] [ WSort [ WAscending "binIMDB_Rating" ], WFrame Nothing (Just 0) ] enc = encoding . position X [ PName "binIMDB_Rating" , pQuant -- the SZero setting is ignored here (presumably as close to 0 anyway) , PScale [ SZero False ] , PTitle "IMDB Rating" ] . position X2 [ PName "binIMDB_Rating_end" ] enc1 = encoding . position Y [ PName "cumulativeCount", pQuant ] enc2 = encoding . position Y [ PName "count", pQuant ] in toVegaLite [ des , dvals , trans [] , enc [] , layer [ asSpec [ enc1 [], mark Bar [] ] , asSpec [ enc2 [], mark Bar [ MColor "yellow", MOpacity 0.5 ] ] ] ] -- advanced16 in elm layered2 :: VegaLite layered2 = let desc = description "Plot showing average data with raw values in the background." dvals = dataFromUrl "https://vega.github.io/vega-lite/data/stocks.csv" [] trans = transform . filter (FExpr "datum.symbol === 'GOOG'") encRaw = encoding . position X [ PName "date", pTemporal, PTimeUnit Year ] . position Y [ PName "price", pQuant ] encAv = encoding . position X [ PName "date", pTemporal, PTimeUnit Year ] . position Y [ PName "price", PAggregate Mean, pQuant ] specRaw = asSpec [ encRaw [], mark Point [ MOpacity 0.3 ] ] specAv = asSpec [ encAv [], mark Line [] ] in toVegaLite [ desc, dvals, trans [], layer [ specRaw, specAv ] ] -- advanced17 in elm layered3 :: VegaLite layered3 = let desc = description "Plot showing a 30 day rolling average with raw values in the background." dvals = dataFromUrl "https://vega.github.io/vega-lite/data/seattle-weather.csv" [] trans = transform . window [ ( [ WAggregateOp Mean, WField "temp_max" ], "rollingMean" ) ] [ WFrame (Just (-15)) (Just 15) ] encRaw = encoding . position X [ PName "date", PTitle "Date", pTemporal ] . position Y [ PName "temp_max", PTitle "Maximum temperature", pQuant ] encAv = encoding . position X [ PName "date", pTemporal ] . position Y [ PName "rollingMean", pQuant ] specRaw = asSpec [ encRaw [], mark Point [ MOpacity 0.3 ] ] specAv = asSpec [ encAv [], mark Line [ MColor "red", MSize 3 ] ] in toVegaLite [ desc, width 400, height 300, dvals, trans [], layer [ specRaw, specAv ] ] -- advanced18 in elm benchmark :: VegaLite benchmark = let desc = description "Line chart to show benchmarking results." toData :: [ Double ] -> Value toData = toJSON falconData = [ 16.81999969482422, 19.759998321533203, 16.079999923706055, 19.579999923706055, 16.420000076293945, 16.200000762939453, 16.020000457763672, 15.9399995803833, 16.280000686645508, 16.119998931884766, 16.15999984741211, 16.119998931884766, 16.139999389648438, 16.100000381469727, 16.200000762939453, 16.260000228881836, 19.35999870300293, 19.700000762939453, 15.9399995803833, 19.139999389648438, 16.200000762939453, 16.119998931884766, 19.520000457763672, 19.700000762939453, 16.200000762939453, 20.979999542236328, 16.299999237060547, 16.420000076293945, 16.81999969482422, 16.5, 16.560001373291016, 16.18000030517578, 16.079999923706055, 16.239999771118164, 16.040000915527344, 16.299999237060547, 19.399999618530273, 15.699999809265137, 16.239999771118164, 15.920000076293945, 16.259998321533203, 16.219999313354492, 16.520000457763672, 16.459999084472656, 16.360000610351563, 15.719999313354492, 16.060001373291016, 15.960000991821289, 16.479999542236328, 16.600000381469727, 16.240001678466797, 16.940000534057617, 16.220001220703125, 15.959999084472656, 15.899999618530273, 16.479999542236328, 16.31999969482422, 15.75999927520752, 15.999998092651367, 16.18000030517578, 16.219999313354492, 15.800000190734863, 16.139999389648438, 16.299999237060547, 16.360000610351563, 16.260000228881836, 15.959999084472656, 15.9399995803833, 16.53999900817871, 16.139999389648438, 16.259998321533203, 16.200000762939453, 15.899999618530273, 16.079999923706055, 16.079999923706055, 15.699999809265137, 15.660000801086426, 16.139999389648438, 23.100000381469727, 16.600000381469727, 16.420000076293945, 16.020000457763672, 15.619999885559082, 16.35999870300293, 15.719999313354492, 15.920001029968262, 15.5600004196167, 16.34000015258789, 22.82000160217285, 15.660000801086426, 15.5600004196167, 16, 16, 15.819999694824219, 16.399999618530273, 16.46000099182129, 16.059999465942383, 16.239999771118164, 15.800000190734863, 16.15999984741211, 16.360000610351563, 19.700000762939453, 16.10000228881836, 16.139999389648438, 15.819999694824219, 16.439998626708984, 16.139999389648438, 16.020000457763672, 15.860000610351563, 16.059999465942383, 16.020000457763672, 15.920000076293945, 15.819999694824219, 16.579999923706055, 15.880000114440918, 16.579999923706055, 15.699999809265137, 19.380001068115234, 19.239999771118164, 16, 15.980000495910645, 15.959999084472656, 16.200000762939453, 15.980000495910645, 16.34000015258789, 16.31999969482422, 16.260000228881836, 15.920000076293945, 15.540000915527344, 16.139999389648438, 16.459999084472656, 16.34000015258789, 15.819999694824219, 19.719999313354492, 15.75999927520752, 16.499998092651367, 15.719999313354492, 16.079999923706055, 16.439998626708984, 16.200000762939453, 15.959999084472656, 16, 16.100000381469727, 19.31999969482422, 16.100000381469727, 16.18000030517578, 15.959999084472656, 22.639999389648438, 15.899999618530273, 16.279998779296875, 16.100000381469727, 15.920000076293945, 16.079999923706055, 16.260000228881836, 15.899999618530273, 15.820001602172852, 15.699999809265137, 15.979998588562012, 16.380001068115234, 16.040000915527344, 19.420000076293945, 15.9399995803833, 16.15999984741211, 15.960000991821289, 16.259998321533203, 15.780000686645508, 15.880000114440918, 15.980000495910645, 16.060001373291016, 16.119998931884766, 23.020000457763672, 15.619999885559082, 15.920000076293945, 16.060001373291016, 14.780000686645508, 16.260000228881836, 19.520000457763672, 16.31999969482422, 16.600000381469727, 16.219999313354492, 19.740001678466797, 19.46000099182129, 15.940000534057617, 15.839999198913574, 16.100000381469727, 16.46000099182129, 16.17999839782715, 16.100000381469727, 15.9399995803833, 16.060001373291016, 15.860000610351563, 15.819999694824219, 16.03999900817871, 16.17999839782715, 15.819999694824219, 17.299999237060547, 15.9399995803833, 15.739999771118164, 15.719999313354492, 15.679998397827148, 15.619999885559082, 15.600000381469727, 16.03999900817871, 15.5, 15.600001335144043, 19.439998626708984, 15.960000991821289, 16.239999771118164, 16.040000915527344, 16.239999771118164 ] squareData = [ 24.200000762939453, 17.899999618530273, 15.800000190734863, 58.400001525878906, 151, 2523.10009765625, 245.3000030517578, 136, 72.30000305175781, 55.70000076293945, 42.400001525878906, 37.70000076293945, 30.100000381469727, 30.100000381469727, 21.799999237060547, 20.600000381469727, 21.799999237060547, 17.600000381469727, 18.200000762939453, 21, 941.7000122070313, 177.39999389648438, 2821.800048828125, 359.20001220703125, 318, 217.10000610351563, 126, 69, 57.79999923706055, 45.29999923706055, 35.599998474121094, 29.100000381469727, 23.799999237060547, 44.20000076293945, 17.700000762939453, 17.700000762939453, 15.699999809265137, 27.799999237060547, 22.799999237060547, 3853.60009765625, 91.5999984741211, 181.39999389648438, 476.29998779296875, 265.8999938964844, 254.60000610351563, 2583.199951171875, 124.80000305175781, 73.19999694824219, 56.400001525878906, 48.70000076293945, 41.599998474121094, 21.100000381469727, 20.299999237060547, 21.299999237060547, 18.299999237060547, 17.100000381469727, 19.5, 828.2000122070313, 162.1999969482422, 217.89999389648438, 205.5, 197.60000610351563, 2249.800048828125, 103.0999984741211, 71.69999694824219, 57.599998474121094, 41.400001525878906, 34.5, 22, 20.5, 21.700000762939453, 18.299999237060547, 17.299999237060547, 19.399999618530273, 666.7999877929688, 214.89999389648438, 212.3000030517578, 125.80000305175781, 67.69999694824219, 56.099998474121094, 45.79999923706055, 38.29999923706055, 33, 35.400001525878906, 22.700000762939453, 19.399999618530273, 19.899999618530273, 24.100000381469727, 19.299999237060547, 21.299999237060547, 3508.699951171875, 204.10000610351563, 125.4000015258789, 65.30000305175781, 60.79999923706055, 44.099998474121094, 36.29999923706055, 30.5, 28.600000381469727, 16.5, 18.600000381469727, 23.700000762939453, 22.299999237060547, 17.600000381469727, 19.200000762939453, 448.79998779296875, 124.4000015258789, 66.5999984741211, 53.5, 51, 45.20000076293945, 28.399999618530273, 29.200000762939453, 26.700000762939453, 25.899999618530273, 18.100000381469727, 17.600000381469727, 20.100000381469727, 25.200000762939453, 3332, 67.5, 53.599998474121094, 56.599998474121094, 39.900001525878906, 27.600000381469727, 29.600000381469727, 33.5, 17.200000762939453, 18.799999237060547, 25.200000762939453, 16.700000762939453, 16.899999618530273, 240.1999969482422, 52.400001525878906, 42.099998474121094, 33.900001525878906, 28, 28.600000381469727, 17.299999237060547, 20, 21, 22.799999237060547, 16.700000762939453, 19.200000762939453, 175.39999389648438, 43.5, 34.70000076293945, 29.700000762939453, 34.900001525878906, 25.799999237060547, 17.299999237060547, 22.600000381469727, 17.600000381469727, 17.200000762939453, 19.200000762939453, 111.80000305175781, 35.400001525878906, 27.600000381469727, 25.399999618530273, 21.899999618530273, 18.600000381469727, 18.100000381469727, 21.200000762939453, 17.899999618530273, 17, 80.5999984741211, 29.799999237060547, 30.100000381469727, 16, 26.799999237060547, 17.5, 22.299999237060547, 16.799999237060547, 22.399999618530273, 77.4000015258789, 31, 29.700000762939453, 28.700000762939453, 26, 16.899999618530273, 15.800000190734863, 19, 52.599998474121094, 25.200000762939453, 16.700000762939453, 17.899999618530273, 21, 19.799999237060547, 18.799999237060547, 46.5, 17.5, 16.799999237060547, 18.299999237060547, 18.299999237060547, 14.899999618530273, 41, 18.299999237060547, 17.299999237060547, 17, 17.5, 32.29999923706055, 22.600000381469727, 16.600000381469727, 17.899999618530273, 25.600000381469727, 17.5, 20.299999237060547, 25.200000762939453, 18.600000381469727, 17.700000762939453 ] trans = transform . window [ ( [ WOp RowNumber ], "row" ) ] [] . calculateAs "1000/datum.data" "fps" transSquare = trans . calculateAs "'Square Crossfilter (3M)'" "system" transFalcon = trans . calculateAs "'Falcon'" "system" enc = encoding . position X [ PName "row" , pQuant , PAxis [ AxGrid False, AxTitle "Trial" ] , PScale [ SNice (IsNice False) ] ] . position Y [ PName "fps" , pQuant , PAxis [ AxGrid False, AxTitle "Frames per Second (fps)" ] , PScale [ SType ScLog ] ] . color [ MName "system" , mNominal , MLegend [ LOrient LOBottomRight, LTitle "System" ] ] . size [ MNumber 1 ] specFalcon = asSpec [ dataFromSource "falcon" [], transFalcon [], mark Line [] ] specSquare = asSpec [ dataFromSource "square" [], transSquare [], mark Line [] ] in toVegaLite [ desc , width 500 , height 300 , datasets [ ( "falcon", dataFromJson (toData falconData) [] ) , ( "square", dataFromJson (toData squareData) [] ) ] , enc [] , layer [ specFalcon, specSquare ] ]