hvega-0.7.0.0: tests/TransformTests.hs
{-# LANGUAGE CPP #-}
{-# LANGUAGE OverloadedStrings #-}
{-# LANGUAGE QuasiQuotes #-}
module TransformTests (testSpecs) where
import qualified Data.Text as T
import Data.Aeson (Value)
import Data.Aeson.QQ.Simple (aesonQQ)
#if !(MIN_VERSION_base(4, 12, 0))
import Data.Monoid ((<>))
#endif
import Prelude hiding (filter, lookup)
import Graphics.Vega.VegaLite
testSpecs :: [(String, VegaLite)]
testSpecs = [ ("checkordering", checkOrdering)
, ("binempty", binEmpty)
, ("binstep", binStep)
, ("imputemean", imputeMean)
, ("densityplot", densityPlot)
, ("loessplot", loessPlot)
, ("lookupplot", lookupPlot)
, ("pivotplot", pivotPlot)
, ("quantileplot", quantilePlot)
, ("regressionplot", regressionPlot)
, ("flattenplot", flattenPlot)
, ("foldasplot", foldAsPlot)
, ("stackplot", stackPlot)
, ("weatherbymonth", weatherByMonth)
, ("weatherbytwomonths", weatherByTwoMonths)
, ("distances", distances)
, ("aggregates", aggregates)
, ("weathermaxbins", weatherMaxBins)
, ("windowplot", windowPlot)
, ("joinaggregateplot", joinAggregatePlot)
]
cars :: Data
cars = dataFromUrl "https://vega.github.io/vega-lite/data/cars.json" []
movies :: [Format] -> Data
movies = dataFromUrl "https://vega.github.io/vega-lite/data/movies.json"
us :: T.Text -> Data
us feature = dataFromUrl "https://vega.github.io/vega-lite/data/us-10m.json"
[ TopojsonFeature feature ]
pos :: Position -> FieldName -> Measurement -> BuildEncodingSpecs
pos ax n t = position ax [ PName n, PmType t ]
checkOrdering :: VegaLite
checkOrdering =
let trans = transform
-- calculateAs transform first to test that order of transforms is preserved.
. calculateAs "datum.Acceleration" "myAcceleration"
. aggregate [ opAs Mean "myAcceleration" "mean_acceleration" ]
[ "Cylinders" ]
enc = encoding
. pos X "Cylinders" Ordinal
. pos Y "mean_acceleration" Quantitative
in toVegaLite [ cars, trans [], enc [], mark Bar [] ]
binTransform :: [BinProperty] -> VegaLite
binTransform bProps =
let trans = transform
. calculateAs "datum.IMDB_Rating" "rating"
. filter (FExpr "datum.rating != null")
. binAs bProps "rating" "ratingGroup"
enc = encoding
. pos X "ratingGroup" Ordinal
. position Y [ PAggregate Count
, PmType Quantitative
, PTitle "Number of movies" ]
in toVegaLite [ width 400, movies [], trans [], enc [], mark Bar [] ]
binEmpty :: VegaLite
binEmpty = binTransform []
binStep :: VegaLite
binStep = binTransform [ Step 0.25 ]
imputeMean :: VegaLite
imputeMean =
let imputeData =
dataFromColumns []
. dataColumn "A" (Numbers [ 0, 0, 1, 1, 2, 2, 3 ])
. dataColumn "B" (Numbers [ 28, 91, 43, 55, 81, 53, 19 ])
. dataColumn "C" (Numbers [ 0, 1, 0, 1, 0, 1, 0 ])
trans =
transform
. calculateAs "datum.A" "a"
. calculateAs "datum.B" "b"
. calculateAs "datum.C" "c"
. impute "b" "a" [ ImMethod ImMean
, ImGroupBy [ "c" ]
, ImFrame (Just (-2)) (Just 2)
]
enc =
encoding
. pos X "a" Quantitative
. pos Y "b" Quantitative
. color [ MName "c", MmType Nominal ]
lopts = [ MStrokeDash [ 5, 10, 5 ]
, MStrokeOpacity 0.5
, MStrokeWidth 2
]
in toVegaLite [ imputeData [], trans [], enc [], mark Line lopts ]
densityPlot :: VegaLite
densityPlot =
let trans = transform
. filter (FExpr "datum.IMDB_Rating != null")
. density "IMDB_Rating" [ DnBandwidth 0.3 ]
enc = encoding
. position X [ PName "value", PmType Quantitative, PTitle "IMDB Rating" ]
. pos Y "density" Quantitative
aopts = [ MFillOpacity 0.8
, MFill "teal"
, MStroke "firebrick"
, MStrokeWidth 2
, MStrokeOpacity 0.7 ]
in toVegaLite [ width 400, height 400, movies [], trans [], enc [], mark Area aopts ]
loessPlot :: VegaLite
loessPlot =
let trans = transform
. calculateAs "datum.IMDB_Rating" "imdbRating"
. calculateAs "datum.Rotten_Tomatoes_Rating" "rtRating"
. loess "imdbRating" "rtRating" [ LsBandwidth 0.1
, LsAs "tx" "ty" ]
enc1 = encoding
. pos X "Rotten_Tomatoes_Rating" Quantitative
. pos Y "IMDB_Rating" Quantitative
enc2 = encoding
. pos X "tx" Quantitative
. pos Y "ty" Quantitative
pointSpec = asSpec [ enc1 [], mark Point [ MFilled True, MOpacity 0.3 ] ]
trendSpec = asSpec [ trans [], enc2 [], mark Line [ MColor "orange" ] ]
in toVegaLite [ width 300, height 300, movies [], layer [ pointSpec, trendSpec ] ]
lookupPlot :: VegaLite
lookupPlot =
let unemployed = dataFromUrl "https://vega.github.io/vega-lite/data/unemployment.tsv" []
trans = transform
. calculateAs "datum.id" "countyID"
. lookup "countyID" unemployed "id" (LuFields [ "rate" ])
proj = projection [ PrType AlbersUsa ]
enc = encoding
. color [ MName "rate"
, MmType Quantitative
, MScale [ SType ScQuantize, SScheme "category10" [ 10 ] ]
]
in toVegaLite [ width 500, height 300, us "counties", proj, trans [], enc [],
mark Geoshape [] ]
pivotPlot :: VegaLite
pivotPlot =
let temps = dataFromColumns []
. dataColumn "city" (Strings [ "Bristol", "Bristol", "Sheffield", "Sheffield", "Glasgow", "Glasgow" ])
. dataColumn "temp" (Numbers [ 12, 14, 11, 13, 7, 10 ])
. dataColumn "year" (Numbers [ 2017, 2018, 2017, 2018, 2017, 2018 ])
trans =
transform
. calculateAs "datum.year" "Year"
. calculateAs "datum.city" "City"
. calculateAs "datum.temp" "Temperature"
. pivot "Year" "Temperature" [ PiGroupBy [ "City" ] ]
-- 2017 temperatures for the Bristol, Sheffield and Glasgow
enc =
encoding
. pos X "2017" Quantitative
. pos Y "City" Nominal
in toVegaLite [ temps [], trans [], enc [], mark Circle [] ]
quantilePlot :: VegaLite
quantilePlot =
let norm = dataFromUrl "https://vega.github.io/vega-lite/data/normal-2d.json" []
trans =
transform
. quantile "u" [ QtStep 0.01, QtAs "p" "v" ]
. calculateAs "quantileUniform(datum.p)" "unif"
. calculateAs "quantileNormal(datum.p)" "norm"
enc1 = encoding . pos X "unif" Quantitative . pos Y "v" Quantitative
enc2 = encoding . pos X "norm" Quantitative . pos Y "v" Quantitative
in toVegaLite [ norm
, trans []
, hConcat [ asSpec [ enc1 [], mark Point [] ]
, asSpec [ enc2 [], mark Point [] ]
]
]
regressionPlot :: VegaLite
regressionPlot =
let trans =
transform
. calculateAs "datum.IMDB_Rating" "imdbRating"
. calculateAs "datum.Rotten_Tomatoes_Rating" "rtRating"
. regression "imdbRating" "rtRating" [ RgMethod RgPoly
, RgOrder 3
, RgExtent 10 90 ]
enc1 = encoding
. pos X "Rotten_Tomatoes_Rating" Quantitative
. pos Y "IMDB_Rating" Quantitative
enc2 = encoding
. pos X "rtRating" Quantitative
. pos Y "imdbRating" Quantitative
pointSpec = asSpec [ enc1 [], mark Point [ MFilled True, MOpacity 0.3 ] ]
regSpec = asSpec [ trans [], enc2 [], mark Line [ MColor "firebrick" ] ]
in toVegaLite [ width 300, height 300, movies [], layer [ pointSpec, regSpec ] ]
dummyData :: Value
dummyData = [aesonQQ|
[ { "key": "alpha", "foo": [ 1, 2 ], "bar": [ "A", "B" ] }
, { "key": "beta", "foo": [ 3, 4, 5 ], "bar": [ "C", "D" ] }
]
|]
flattenPlot :: VegaLite
flattenPlot =
let dvals = dataFromJson dummyData []
trans = transform . flattenAs [ "foo", "bar" ] [ "quant", "cat" ]
enc = encoding
. pos X "quant" Quantitative
. pos Y "cat" Nominal
. color [ MName "key", MmType Nominal ]
in toVegaLite [ dvals, trans [], mark Circle [], enc [] ]
foldAsPlot :: VegaLite
foldAsPlot =
let dvals = dataFromColumns []
. dataColumn "country" (Strings [ "USA", "Canada" ])
. dataColumn "gold" (Numbers [ 10, 7 ])
. dataColumn "silver" (Numbers [ 20, 26 ])
trans = transform
. calculateAs "datum.gold" "goldMedals"
. calculateAs "datum.silver" "silverMedals"
. foldAs [ "goldMedals", "silverMedals" ] "k" "v"
. calculateAs "datum.k" "year"
. calculateAs "datum.v" "numberOfMedals"
enc = encoding
. column [ FName "year", FmType Nominal ]
. pos X "country" Nominal
. pos Y "numberOfMedals" Quantitative
. color [ MName "country", MmType Nominal, MLegend [] ]
in toVegaLite [ dvals [], trans [], mark Bar [], enc [] ]
stackPlot :: VegaLite
stackPlot =
let trans = transform
. aggregate [ opAs Count "" "count_*" ] [ "Origin", "Cylinders" ]
. stack "count_*"
[]
"stack_count_Origin1"
"stack_count_Origin2"
[ StOffset StNormalize, StSort [ WAscending "Origin" ] ]
. window
[ ( [ WAggregateOp Min, WField "stack_count_Origin1" ], "x" )
, ( [ WAggregateOp Max, WField "stack_count_Origin2" ], "x2" )
]
[ WFrame Nothing Nothing, WGroupBy [ "Origin" ] ]
. stack "count_*"
[ "Origin" ]
"y"
"y2"
[ StOffset StNormalize, StSort [ WAscending "Cylinders" ] ]
enc = encoding
. position X [ PName "x", PmType Quantitative, PAxis [] ]
. position X2 [ PName "x2" ]
. position Y [ PName "y", PmType Quantitative, PAxis [] ]
. position Y2 [ PName "y2" ]
. color [ MName "Origin", MmType Nominal ]
. opacity [ MName "Cylinders", MmType Quantitative, MLegend [] ]
. tooltips
[ [ TName "Origin", TmType Nominal ]
, [ TName "Cylinders", TmType Quantitative ]
]
in toVegaLite [ cars, trans [], enc [], mark Rect [] ]
weather :: TimeUnit -> FieldName -> VegaLite
weather tunit field =
let weatherData = dataFromUrl "https://vega.github.io/vega-lite/data/seattle-weather.csv"
[ Parse [ ( "date", FoDate "%Y/%m/%d" ) ] ]
trans = transform
. calculateAs "datum.date" "sampleDate"
. calculateAs "datum.temp_max" "maxTemp"
. timeUnitAs tunit "sampleDate" field
enc = encoding
. position X [ PName field, PmType Temporal, PAxis [ AxFormat "%b" ] ]
. position Y [ PName "maxTemp", PmType Quantitative, PAggregate Max ]
in toVegaLite [ width 400
, weatherData
, trans []
, enc []
, mark Line [ MPoint (PMMarker [ MFill "black" ]) ]
]
weatherByMonth :: VegaLite
weatherByMonth = weather Month "month"
weatherByTwoMonths :: VegaLite
weatherByTwoMonths = weather (TUStep 2 Month) "bimonth"
weatherMaxBins :: VegaLite
weatherMaxBins = weather (TUMaxBins 3) "tbin"
distances :: VegaLite
distances =
let dateTime mnt = "Sun, 01 Jan 2012 00:0" <> T.pack (show mnt) <> ":00"
dates = map dateTime [1 :: Int .. 15]
dvals = dataFromColumns []
. dataColumn "date" (Strings dates)
. dataColumn "distance" (Numbers [ 1, 1, 2, 1, 4, 2, 5, 2, 6, 4, 1, 1, 3, 0, 2, 3 ])
enc = encoding
. position X [ PName "date"
, PmType Temporal
, PTimeUnit (TUMaxBins 15) ]
. position Y [ PName "distance"
, PmType Quantitative
, PAggregate Sum ]
in toVegaLite [ dvals [], enc [], mark Bar [] ]
aggregates :: VegaLite
aggregates =
let dvals = dataFromColumns []
. dataColumn "xs" (Strings ["A", "A", "B", "A", "B", "B", "C", "C"])
. dataColumn "ys" (Numbers [1, 2, 2, 2, 3, 2, 4, 8])
enc op = encoding (position Y [ PName "ys"
, PAxis [AxTitle "Y"]
, PmType Quantitative
, PAggregate op ] [])
lyr1 = asSpec [ enc Sum, mark Bar [MStroke "black"] ]
lyr2 = asSpec [ enc Product, mark Line [MColor "brown"] ]
lyr3 = asSpec [ enc Max, mark Point [MColor "orange"] ]
lyr4 = asSpec [ enc Distinct, mark Square [MColor "cyan"] ]
axOpts = PAxis [AxTitle "X", AxLabelAngle 0]
in toVegaLite [ dvals []
, encoding (position X [PName "xs", axOpts, PmType Ordinal] [])
, layer [lyr1, lyr2, lyr3, lyr4]
]
activityData :: Data
activityData =
dataFromColumns []
. dataColumn "Activity" (Strings [ "Sleeping", "Eating", "TV", "Work", "Exercise" ])
. dataColumn "Time" (Numbers [ 8, 2, 4, 8, 2 ])
$ []
windowPlot :: VegaLite
windowPlot =
let 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, PTitle "% of total time" ]
. pos Y "Activity" Nominal
in toVegaLite
[ heightStep 12
, activityData
, trans []
, mark Bar []
, enc []
]
joinAggregatePlot :: VegaLite
joinAggregatePlot =
let trans = transform
. joinAggregate [ opAs Sum "Time" "TotalTime" ] []
. calculateAs "datum.Time/datum.TotalTime * 100" "PercentOfTotal"
enc = encoding
. position X [ PName "PercentOfTotal", PmType Quantitative, PTitle "% of total time" ]
. pos Y "Activity" Nominal
in toVegaLite
[ heightStep 12
, activityData
, trans []
, mark Bar []
, enc []
]