private Action makeLineChart(Field x, Field y) {
// If there is close to one data point per x coordinate just use lines
if (x.numericProperty("unique") > 0.95 * x.numericProperty("validNumeric"))
return get("line").apply(x, y);
// Otherwise show lines and points
return get("lineWithPoints").apply(x, y);
}
private boolean goodForWordle(Field f) {
if (!f.preferCategorical()) return false;
if (f.numericProperty("unique") > 100 || f.numericProperty("unique") < 7) return false;
// Too long names are not good
for (Object c : f.categories()) if (c.toString().length() > 20) return false;
return true;
}
private boolean allShowExtent(Field[] fields) {
// Categorical and numeric fields both show elements as extents on the axis
for (Field field : fields) {
if (field.isNumeric() && !field.isBinned()) return false;
}
return true;
}
private String asString(NumericScale details, Field field) {
String s =
details.type + " : " + field.format(details.min) + " " + field.format(details.max) + " : ";
for (int i = 0; i < details.divisions.length; i++)
s += "|" + field.format(details.divisions[i]);
return s;
}
@Test
public void testMakingField() {
Field f = Data.makeColumnField("a", "label", new Object[] {"1", "2", "a", "2", null, 1});
assertEquals("a", f.name);
assertEquals("label", f.label);
assertEquals(6, f.rowCount());
assertEquals(5, f.valid());
}
private void setLocations(
ElementDefinition.ElementDimensionDefinition dim,
String dimName,
Field[] fields,
boolean categorical) {
String scaleName = "scale_" + dimName;
if (fields.length == 0) {
// There are no fields -- we have a notional [0,1] extent, so use the center of that
dim.center = "function() { return " + scaleName + "(0.5) }";
dim.left = "function() { return " + scaleName + "(0) }";
dim.right = "function() { return " + scaleName + "(1) }";
} else if (fields.length == 1) {
Field field = fields[0]; // The single field
String dataFunction = D3Util.writeCall(field); // A call to that field using the datum 'd'
if (isRange(field)) {
// This is a range field, but we have not been asked to show both ends,
// so we use the difference between the top and bottom
dim.center = "function(d) { return " + scaleName + "(" + dataFunction + ".extent()) }";
// Left and Right are not defined
} else if (field.isBinned() && !categorical) {
// A Binned value on a non-categorical axes
dim.center = "function(d) { return " + scaleName + "(" + dataFunction + ".mid) }";
dim.left = "function(d) { return " + scaleName + "(" + dataFunction + ".low) }";
dim.right = "function(d) { return " + scaleName + "(" + dataFunction + ".high) }";
} else {
// Nothing unusual -- just define the center
dim.center = "function(d) { return " + scaleName + "(" + dataFunction + ") }";
}
} else {
// The dimension contains two fields: a range
String lowDataFunc =
D3Util.writeCall(fields[0]); // A call to the low field using the datum 'd'
String highDataFunc =
D3Util.writeCall(fields[1]); // A call to the high field using the datum 'd'
// When one of the fields is a range, use the outermost value of that
if (isRange(fields[0])) lowDataFunc += ".low";
if (isRange(fields[1])) highDataFunc += ".high";
dim.left = "function(d) { return " + scaleName + "(" + lowDataFunc + ") }";
dim.right = "function(d) { return " + scaleName + "(" + highDataFunc + ") }";
dim.center =
"function(d) { return "
+ scaleName
+ "( ("
+ highDataFunc
+ " + "
+ lowDataFunc
+ " )/2) }";
}
}
@Test
public void testDateConversionForBadDates() {
Object[] data = new Object[1000];
for (int i = 0; i < data.length; i++) data[i] = i % 10 == 0 ? "name" : "-";
Field f = Data.makeColumnField("a", "label", data);
Field g = Data.toDate(f);
assertNotEquals(f, g);
assertEquals("a", g.name);
assertEquals("label", g.label);
assertEquals(1000, g.rowCount());
assertEquals(0, g.valid());
}
private void setProperties(String f, Field to, Field src) {
// Nothing to set for a list
if (f.equals("list")) {
// Need to keep the date format
to.copyProperties(src, "dateFormat");
to.set("list", true);
} else if (f.equals("count") || f.equals("percent") || f.equals("valid") || f.equals("unique"))
// These are numeric, but do not preserve properties
to.setNumeric();
else
// All properties copy over
Fields.copyBaseProperties(src, to);
}
@Test
public void testNumericConversion() {
Field f = Data.makeColumnField("a", "label", new Object[] {"1", "2", "a", "2", null, 1});
Field g = Data.toNumeric(f);
assertNotEquals(f, g);
assertEquals("a", g.name);
assertEquals("label", g.label);
assertEquals(6, g.rowCount());
assertEquals(4, g.valid());
// Should not change
assertEquals(g, Data.toNumeric(g));
}
@Test
public void testTiedValues() {
Field f = Data.makeColumnField("a", "label", new Object[] {200, 100, 400, 500, 100});
Field g = Data.makeColumnField("a", "label", new Object[] {200, 100, 400, 500, 600});
Field h = Data.makeColumnField("a", "label", new Object[] {200, 100, 400, 100, 200});
Field i = Data.makeColumnField("a", "label", new Object[] {100, 100, 100, 500, 500});
Field j =
Data.makeColumnField("a", "label", new Object[] {1, 2, 3, 4, 1, 2, 3, 4, 1, 2, 3, 4, 3, 4});
assertEquals(100, f.numericProperty("mode"), 0.01);
assertEquals(400, g.numericProperty("mode"), 0.01);
assertEquals(100, h.numericProperty("mode"), 0.01);
assertEquals(100, i.numericProperty("mode"), 0.01);
assertEquals(3, j.numericProperty("mode"), 0.01);
}
@Test
public void testPadding() {
Field a =
Data.toNumeric(Data.makeColumnField("a", "label", new Object[] {100, 200, 120, 200, 3100}));
a.set("transform", "linear");
Assert.assertEquals(
"linear : 70 3130 : |500|1000|1500|2000|2500|3000",
asString(
NumericScale.makeLinearScale(a, false, 0.0, new double[] {0.01, 0.01}, 5, false), a));
// Does not pad past zero
Assert.assertEquals(
"linear : -200 3400 : |0|1000|2000|3000",
asString(
NumericScale.makeLinearScale(a, false, 0.0, new double[] {0.1, 0.1}, 5, false), a));
}
@Test
public void testProperties() {
Field f = Data.makeColumnField("a", "label", new Object[0]);
assertEquals(null, f.property("xyz"));
f.set("xyz", "12");
assertEquals("12", f.stringProperty("xyz"));
assertEquals(12.00001, f.numericProperty("xyz"), 0.001);
f.set("abc", false);
assertEquals(Boolean.FALSE, f.property("abc"));
f.set("abc", true);
assertEquals(Boolean.TRUE, f.property("abc"));
}
@Test
public void testPermute() {
Field base = Data.makeIndexingField("a", "b", 10);
Field a = Data.permute(base, new int[] {0, 0, 0, 1, 1, 2, 2, 3}, false);
assertEquals(true, a.hasProperty("numeric"));
assertEquals("a", a.name);
assertEquals("b", a.label);
assertEquals(8, a.rowCount());
assertEquals(8, a.getNumericProperty("valid"), 0.001);
assertEquals(8, a.getNumericProperty("validNumeric"), 0.001);
assertEquals(4, a.getNumericProperty("unique"), 0.001);
assertEquals((1 + 1 + 1 + 2 + 2 + 3 + 3 + 4) / 8.0, a.getNumericProperty("mean"), 0.001);
assertEquals(1, a.value(1));
assertEquals(1, a.value(2));
assertEquals(4, a.value(7));
assertEquals("1, 2, 3, 4", Data.join(a.categories()));
}
Field(String name, String label, Provider provider, Field base) {
this.name = name;
this.label = label == null ? name : label;
this.provider = provider;
// Information is provided in the base field
if (base != null) {
if (provider == null) {
// Ensure that the base field has everything calculated because we cannot calculate lazily
// later
base.makeNominalStats();
base.makeNumericStats();
base.makeDateStats();
}
copyPropertiesFrom(base);
}
}
private Action chooseBivariate(Field x, Field y) {
// Dates need line charts
if (x.isDate() && !y.isDate() && !y.preferCategorical()) return makeLineChart(x, y);
if (y.isDate() && !x.isDate() && !x.preferCategorical()) return makeLineChart(y, x);
// Heatmaps if either is categorical
if (x.preferCategorical() || y.preferCategorical())
return get("heatmap").apply(orderByCoarseness(x, y));
// Default to scatter
return get("scatter").apply(orderByCoarseness(x, y));
}
@Test
@JSTranslation(ignore = true)
public void testJavaDates() {
long feb_1_1970 = 60907593600000L;
Field f =
Data.makeColumnField(
"test",
"dates",
new Object[] {
new Date(feb_1_1970),
new Date(feb_1_1970 + 1000 * 60 * 60 * 24),
new Date(feb_1_1970 + 2 * 1000 * 60 * 60 * 24),
new Date(feb_1_1970 + 3 * 1000 * 60 * 60 * 24),
new Date(feb_1_1970 + 4 * 1000 * 60 * 60 * 24),
new Date(feb_1_1970 + 5 * 1000 * 60 * 60 * 24),
new Date(feb_1_1970 + 6 * 1000 * 60 * 60 * 24),
new Date(feb_1_1970 + 7 * 1000 * 60 * 60 * 24),
new Date(feb_1_1970 + 8 * 1000 * 60 * 60 * 24),
new Date(feb_1_1970 + 9 * 1000 * 60 * 60 * 24),
new Date(feb_1_1970 + 10 * 1000 * 60 * 60 * 24),
new Date(feb_1_1970 + 20 * 1000 * 60 * 60 * 24)
});
f.set("numeric", true);
f.set("date", true);
assertTrue(f.isDate());
assertEquals("Feb 1 00:00", f.valueFormatted(0));
assertEquals("Feb 2 00:00", f.valueFormatted(1));
assertEquals("Feb 3 00:00", f.valueFormatted(2));
assertEquals("Feb 11 00:00", f.valueFormatted(10));
assertEquals("Feb 21 00:00", f.valueFormatted(11));
}
@Test
public void testDateConversion() {
Field f =
Data.makeColumnField(
"a",
"label",
new Object[] {"Jan 4, 1980", "Jan 4, 1988", "a", "2", null, "Jan 9, 1978"});
Field g = Data.toDate(f, null);
assertNotEquals(f, g);
assertEquals("a", g.name);
assertEquals("label", g.label);
assertEquals(6, g.rowCount());
assertEquals(3, g.valid());
// Should not change
assertEquals(g, Data.toDate(g, null));
}
@Test
public void testLinearScale() {
double[] pad = {0, 0};
Field a = Data.makeColumnField("a", "label", new Object[] {2, 4, 7, 120, 45, 120, 200, 3345});
a = Data.toNumeric(a);
a.set("transform", "linear");
Assert.assertEquals(
"linear : 2 3345 : |500|1000|1500|2000|2500|3000",
asString(NumericScale.makeLinearScale(a, false, 0.0, pad, 5, false), a));
Assert.assertEquals(
"linear : 2 3345 : |200|400|600|800|1000|1200|1400|1600|1800|2000|2200|2400|2600|2800|3000|3200",
asString(NumericScale.makeLinearScale(a, false, 0.0, pad, 15, false), a));
Assert.assertEquals(
"linear : 2 3345 : |1000|2000|3000",
asString(NumericScale.makeLinearScale(a, false, 0.0, pad, 2, false), a));
Assert.assertEquals(
"linear : 2 3345 : |2000",
asString(NumericScale.makeLinearScale(a, false, 0.0, pad, 1, false), a));
Assert.assertEquals(
"linear : 0 3345 : |0|500|1000|1500|2000|2500|3000",
asString(NumericScale.makeLinearScale(a, false, 0.01, pad, 7, false), a));
Assert.assertEquals(
"linear : 0 3500 : |0|500|1000|1500|2000|2500|3000|3500",
asString(NumericScale.makeLinearScale(a, true, 0.0, pad, 7, false), a));
a = Data.makeColumnField("a", "label", new Object[] {-22.2, -22.201, -22.9, -22.7});
a = Data.toNumeric(a);
Assert.assertEquals(
"linear : -23 -22 : |-23|-22.8|-22.6|-22.4|-22.2|-22",
asString(Auto.makeNumericScale(a, true, pad, 0.0, 5, false), a));
Assert.assertEquals(
"linear : -22.9 -22.2 : |-22.8|-22.6|-22.4|-22.2",
asString(NumericScale.makeLinearScale(a, false, 0.0, pad, 4, false), a));
Assert.assertEquals(
"linear : -23 -22 : |-23|-22.5|-22",
asString(Auto.makeNumericScale(a, true, pad, 0.0, 4, false), a));
Assert.assertEquals(
"linear : -23 -22 : |-23|-22.5|-22",
asString(Auto.makeNumericScale(a, true, pad, 0.95, 4, false), a));
Assert.assertEquals(
"linear : -30 0 : |-30|-20|-10|0",
asString(NumericScale.makeLinearScale(a, true, 0.99, pad, 4, false), a));
}
@Test
public void testIndexing() {
Field a = Data.makeIndexingField("a", "b", 30);
assertEquals("a", a.name);
assertEquals("b", a.label);
assertEquals(30, a.rowCount());
assertEquals(30, a.getNumericProperty("valid"), 0.001);
assertEquals(30, a.getNumericProperty("validNumeric"), 0.001);
assertEquals(30, a.getNumericProperty("unique"), 0.001);
assertEquals(15.5, a.getNumericProperty("mean"), 0.001);
assertEquals(3, a.value(2));
assertEquals(8, a.value(7));
assertEquals(true, a.hasProperty("numeric"));
}
@Test
public void testConstant() {
Field a = Data.makeConstantField("a", "b", "foo", 50);
assertEquals("a", a.name);
assertEquals("b", a.label);
assertEquals(50, a.rowCount());
assertEquals(50, a.getNumericProperty("valid"), 0.001);
assertEquals(0, a.getNumericProperty("validNumeric"), 0.001);
assertEquals(1, a.getNumericProperty("unique"), 0.001);
assertEquals("foo", a.value(2));
assertEquals(false, a.hasProperty("numeric"));
}
@Test
public void testDateScaleHoursOrLess() {
Field a;
double[] pad = {0, 0};
a = Data.toDate(Data.makeColumnField("a", "label", new Object[] {"09:45:22", "09:45:24"}));
Assert.assertEquals(
"date : 09:45:22 09:45:24 : |09:45:22|09:45:23|09:45:24",
asString(NumericScale.makeDateScale(a, false, pad, 5), a));
a = Data.toDate(Data.makeColumnField("a", "label", new Object[] {"09:45:22", "09:45:54"}));
assertEquals(0.40650462963, a.min(), 1e-9);
assertEquals(0.406875, a.max(), 1e-9);
Assert.assertEquals(
"date : 09:45:22 09:45:54 : |09:45:30|09:45:40|09:45:50",
asString(NumericScale.makeDateScale(a, false, pad, 5), a));
a = Data.toDate(Data.makeColumnField("a", "label", new Object[] {"09:45:22", "09:45:54"}));
Assert.assertEquals(
"date : 09:45:20 09:46:00 : |09:45:20|09:45:30|09:45:40|09:45:50|09:46:00",
asString(NumericScale.makeDateScale(a, true, pad, 5), a));
a = Data.toDate(Data.makeColumnField("a", "label", new Object[] {"09:45:22", "10:01:54"}));
Assert.assertEquals(
"date : 09:45 10:05 : |09:45|09:50|09:55|10:00|10:05",
asString(NumericScale.makeDateScale(a, true, pad, 5), a));
a = Data.toDate(Data.makeColumnField("a", "label", new Object[] {"09:45:22", "12:01:54"}));
Assert.assertEquals(
"date : 09:30 12:30 : |09:30|10:00|10:30|11:00|11:30|12:00|12:30",
asString(NumericScale.makeDateScale(a, true, pad, 5), a));
a = Data.toDate(Data.makeColumnField("a", "label", new Object[] {"09:45:22", "23:01:54"}));
Assert.assertEquals(
"date : 09:00 00:00 : |09:00|12:00|15:00|18:00|21:00|00:00",
asString(NumericScale.makeDateScale(a, true, pad, 5), a));
}
@Test
public void testDate() {
Field a;
a =
Data.toDate(
Data.makeColumnField(
"a", "label", new Object[] {"January 1, 1970", "January 1, 1970 00:00:01"}));
assertEquals(0.0, a.min(), 1e-9);
assertEquals(1.0 / 3600 / 24, a.max(), 1e-9);
assertEquals(DateUnit.second, a.property("dateUnit"));
a =
Data.toDate(
Data.makeColumnField(
"a", "label", new Object[] {"January 1, 1970", "January 1, 1970 00:04:00"}));
assertEquals(0.0, a.min(), 1e-9);
assertEquals(4.0 / 60 / 24, a.max(), 1e-9);
assertEquals(DateUnit.minute, a.property("dateUnit"));
a =
Data.toDate(
Data.makeColumnField(
"a", "label", new Object[] {"January 1, 1970", "January 1, 1970 00:06:00"}));
assertEquals(0.0, a.min(), 1e-9);
assertEquals(6.0 / 60 / 24, a.max(), 1e-9);
assertEquals(DateUnit.minute, a.property("dateUnit"));
a =
Data.toDate(
Data.makeColumnField(
"a", "label", new Object[] {"January 1, 1970", "January 1, 1970 00:45:00"}));
assertEquals(0.0, a.min(), 1e-9);
assertEquals(45.0 / 60 / 24, a.max(), 1e-9);
assertEquals(DateUnit.minute, a.property("dateUnit"));
a =
Data.toDate(
Data.makeColumnField(
"a", "label", new Object[] {"January 1, 1970", "January 1, 1970 04:59:00"}));
assertEquals(0.0, a.min(), 1e-9);
assertEquals(DateUnit.hour, a.property("dateUnit"));
a =
Data.toDate(
Data.makeColumnField(
"a", "label", new Object[] {"January 1, 1970", "January 1, 1970 05:00:00"}));
assertEquals(0.0, a.min(), 1e-9);
assertEquals(5.0 / 24, a.max(), 1e-9);
assertEquals(DateUnit.hour, a.property("dateUnit"));
a =
Data.toDate(
Data.makeColumnField(
"a", "label", new Object[] {"January 1, 1970", "January 4, 1970"}));
assertEquals(0.0, a.min(), 1e-9);
assertEquals(3.0, a.max(), 1e-9);
assertEquals(DateUnit.day, a.property("dateUnit"));
// Note that only three days are needed to get days as a unit -- this an exception to the usual
// "5 ticks" rule
a =
Data.toDate(
Data.makeColumnField(
"a", "label", new Object[] {"January 10, 1970", "January 2, 1970"}));
assertEquals(1.0, a.min(), 1e-9);
assertEquals(9.0, a.max(), 1e-9);
assertEquals(DateUnit.day, a.property("dateUnit"));
a =
Data.toDate(
Data.makeColumnField("a", "label", new Object[] {"January 10, 1970", "March 2, 1970"}));
assertEquals(9.0, a.min(), 1e-9);
assertEquals(60.0, a.max(), 1e-9);
assertEquals(DateUnit.week, a.property("dateUnit"));
a =
Data.toDate(
Data.makeColumnField(
"a", "label", new Object[] {"January 10, 1970", "December 31, 1970"}));
assertEquals(9.0, a.min(), 1e-9);
assertEquals(364.0, a.max(), 1e-9);
assertEquals(DateUnit.quarter, a.property("dateUnit"));
a =
Data.toDate(
Data.makeColumnField(
"a", "label", new Object[] {"January 10, 1970", "January 10, 1972"}));
assertEquals(9.0, a.min(), 1e-9);
assertEquals(2 * 365 + 9, a.max(), 1e-9);
assertEquals(DateUnit.quarter, a.property("dateUnit"));
a =
Data.toDate(
Data.makeColumnField(
"a", "label", new Object[] {"January 10, 1970", "January 10, 1974"}));
assertEquals(9.0, a.min(), 1e-9);
assertEquals(4 * 365 + 1 + 9, a.max(), 1e-9);
assertEquals(DateUnit.year, a.property("dateUnit"));
a =
Data.toDate(
Data.makeColumnField(
"a", "label", new Object[] {"January 10, 1970", "January 10, 1976"}));
assertEquals(9.0, a.min(), 1e-9);
assertEquals(6 * 365 + 1 + 9, a.max(), 1e-9);
assertEquals(DateUnit.year, a.property("dateUnit"));
a =
Data.toDate(
Data.makeColumnField(
"a", "label", new Object[] {"January 10, 1970", "January 10, 2030"}));
assertEquals(9.0, a.min(), 1e-9);
assertEquals(60 * 365 + 15 + 9, a.max(), 1e-9);
assertEquals(DateUnit.decade, a.property("dateUnit"));
a =
Data.toDate(
Data.toDate(
Data.makeColumnField("a", "label", new Object[] {"09:45:22", "09:45:24"}), null));
assertEquals((9 + 45 / 60.0 + 22 / 3600.0) / 24.0, a.min(), 1e-9);
assertEquals((9 + 45 / 60.0 + 24 / 3600.0) / 24.0, a.max(), 1e-9);
assertEquals(DateUnit.second, a.property("dateUnit"));
}
@Test
public void testChooseTransform() {
Field a = Data.makeColumnField("a", "label", new Object[] {100, 200, 300, 400, 500, 600});
Field b = Data.makeColumnField("b", "label", new Object[] {"a", "b", "c"});
Field c = Data.makeColumnField("d", "label", new Object[] {1, 2, 2, 2, 2, 2, 2, 3});
Field d = Data.makeColumnField("d", "label", new Object[] {1, 1, 1, 1, 1, 2, 2, 2, 5, 10});
Field e =
Data.makeColumnField("e", "label", new Object[] {1, 1, 1, 1, 1, 2, 2, 2, 5, 10, 100, 1000});
Field f =
Data.makeColumnField("f", "label", new Object[] {0, 1, 1, 1, 1, 2, 2, 2, 5, 10, 100, 1000});
Field g =
Data.makeColumnField(
"g", "label", new Object[] {-1, 1, 1, 1, 1, 2, 2, 2, 5, 10, 100, 1000});
Auto.setTransform(a);
Auto.setTransform(b);
Auto.setTransform(c);
Auto.setTransform(d);
Auto.setTransform(e);
Auto.setTransform(f);
Auto.setTransform(g);
Assert.assertEquals("linear", a.stringProperty("transform"));
Assert.assertEquals("linear", b.stringProperty("transform"));
Assert.assertEquals("linear", c.stringProperty("transform"));
Assert.assertEquals("root", d.stringProperty("transform"));
Assert.assertEquals("log", e.stringProperty("transform"));
Assert.assertEquals("root", f.stringProperty("transform"));
Assert.assertEquals("linear", g.stringProperty("transform"));
}
@Test
public void testFieldStats() {
Field uniform = Data.makeColumnField("a", "label", new Object[] {100, 200, 300, 400, 500, 600});
Field uniformWithMissing =
Data.makeColumnField("a", "label", new Object[] {100, 200, null, 300, "a", 400, 500, 600});
Field peak = Data.makeColumnField("b", "label", new Object[] {1, 2, 2, 2, 2, 2, 2, 3});
Field skew = Data.makeColumnField("c", "label", new Object[] {1, 1, 1, 1, 1, 2, 2, 2, 5, 10});
Field a =
Data.makeColumnField("f", "label", new Object[] {0, 1, 1, 1, 1, 2, 2, 2, 5, 10, 100, 1000});
Field b = Data.makeColumnField("f", "label", new Object[] {10, 20, 30, 40, 22, 50, 60});
uniform = uniform.dropData();
uniformWithMissing = uniformWithMissing.dropData();
peak = peak.dropData();
skew = skew.dropData();
a = a.dropData();
b = b.dropData();
// Basics
assertEquals(6, uniform.rowCount());
assertEquals(8, uniformWithMissing.rowCount());
assertEquals(350, uniform.numericProperty("median"), 0.01);
assertEquals(350, uniformWithMissing.numericProperty("median"), 0.01);
assertEquals(2, peak.numericProperty("median"), 0.01);
assertEquals(1.5, skew.numericProperty("median"), 0.01);
assertEquals(2, a.numericProperty("median"), 0.01);
assertEquals(30, b.numericProperty("median"), 0.01);
assertEquals(200, uniform.numericProperty("q1"), 0.01);
assertEquals(200, uniformWithMissing.numericProperty("q1"), 0.01);
assertEquals(2, peak.numericProperty("q1"), 0.01);
assertEquals(1, skew.numericProperty("q1"), 0.01);
assertEquals(1, a.numericProperty("q1"), 0.01);
assertEquals(21, b.numericProperty("q1"), 0.01);
assertEquals(500, uniform.numericProperty("q3"), 0.01);
assertEquals(500, uniformWithMissing.numericProperty("q3"), 0.01);
assertEquals(2, peak.numericProperty("q3"), 0.01);
assertEquals(2, skew.numericProperty("q3"), 0.01);
assertEquals(7.5, a.numericProperty("q3"), 0.01);
assertEquals(45, b.numericProperty("q3"), 0.01);
assertEquals(100, uniform.numericProperty("granularity"), 0.01);
assertEquals(100, uniformWithMissing.numericProperty("granularity"), 0.01);
assertEquals(1, peak.numericProperty("granularity"), 0.01);
assertEquals(1, skew.numericProperty("granularity"), 0.01);
assertEquals(1, a.numericProperty("granularity"), 0.01);
assertEquals(2, b.numericProperty("granularity"), 0.01);
}
@Test
public void testBasicFieldStats() {
Field f = Data.makeColumnField("a", "label", new Object[] {"1", "2", "a", "2", null, 0});
Field g = Data.makeColumnField("a", "label", new Object[] {100, 200, 400, 500, 600});
Field h = Data.makeColumnField("a", "label", new Object[] {"a", "b", "c", "c"});
Field i = Data.makeIndexingField("a", "label", 20);
Field c = Data.makeConstantField("a", "label", 5.0, 2000);
assertEquals(0, f.min(), 0.01);
assertEquals(2, f.max(), 0.01);
assertEquals(100, g.min(), 0.01);
assertEquals(600, g.max(), 0.01);
assertEquals(null, h.min());
assertEquals(null, h.max());
assertEquals(1, i.min(), 0.01);
assertEquals(20, i.max(), 0.01);
assertEquals(5.0, c.min(), 0.01);
assertEquals(5.0, c.max(), 0.01);
int[] counts = (int[]) h.property("categoryCounts");
Assert.assertEquals(1, counts[0]);
Assert.assertEquals(1, counts[1]);
Assert.assertEquals(2, counts[2]);
assertEquals(4, f.uniqueValuesCount());
assertEquals(5, g.uniqueValuesCount());
assertEquals(3, h.uniqueValuesCount());
assertEquals(20, i.uniqueValuesCount());
assertEquals(1, c.uniqueValuesCount());
}
@Test
public void testAutoConvertDate() {
// Make some dates
Date d1 = Data.asDate("1971-1-3");
Date d2 = Data.asDate("1971-1-9");
Date d3 = Data.asDate("1971-1-31");
Field a = Data.makeColumnField("field", null, new Object[] {d1, d2, d3, null});
Field b = Auto.convert(a);
Assert.assertEquals(true, b.isNumeric());
Assert.assertEquals(true, b.isDate());
// Should ignore one error
a = Data.makeColumnField("field", null, new Object[] {d1, d2, "oops", d3, null, null});
b = Auto.convert(a);
Assert.assertEquals(true, b.isNumeric());
Assert.assertEquals(true, b.isDate());
assertEquals(3, b.valid());
a = Data.makeColumnField("field", null, new Object[] {"1971-1-3", "1971-1-12"});
b = Auto.convert(a);
Assert.assertEquals(true, b.isNumeric());
Assert.assertEquals(true, b.isDate());
assertEquals(2, b.valid());
assertEquals(9.0, b.max() - b.min(), 0.001);
a = Data.makeColumnField("field", null, new Object[] {1970, 1972, 1978});
b = Auto.convert(a);
Assert.assertEquals(true, b.isNumeric());
Assert.assertEquals(true, b.isDate());
assertEquals(3, b.valid());
assertEquals(8 * 365 + 2, b.max() - b.min(), 0.001);
a = Data.makeColumnField("field", null, new Object[] {1970, 1971.5, 1978});
b = Auto.convert(a);
Assert.assertEquals(true, b.isNumeric());
Assert.assertEquals(false, b.isDate());
assertEquals(3, b.valid());
assertEquals(8, b.max() - b.min(), 0.001);
}
@Test
public void testEmptyField() {
Field f = Data.makeColumnField("a", "label", new Object[0]);
assertEquals(0, f.rowCount());
}
private boolean isRange(Field field) {
String s = field.stringProperty("summary");
return s != null && (s.equals("iqr") || s.equals("range"));
}
@Test
public void testMomentFieldStats() {
Field uniform = Data.makeColumnField("a", "label", new Object[] {100, 200, 300, 400, 500, 600});
Field uniformWithMissing =
Data.makeColumnField("a", "label", new Object[] {100, 200, null, 300, "a", 400, 500, 600});
Field peak = Data.makeColumnField("b", "label", new Object[] {1, 2, 2, 2, 2, 2, 2, 3});
Field skew = Data.makeColumnField("c", "label", new Object[] {1, 1, 1, 1, 1, 2, 2, 2, 5, 10});
assertEquals(350, uniform.numericProperty("mean"), 0.01);
assertEquals(350, uniformWithMissing.numericProperty("mean"), 0.01);
assertEquals(2, peak.numericProperty("mean"), 0.01);
assertEquals(2.6, skew.numericProperty("mean"), 0.01);
assertEquals(187.08, uniform.numericProperty("stddev"), 0.01);
assertEquals(187.08, uniformWithMissing.numericProperty("stddev"), 0.01);
assertEquals(0.534, peak.numericProperty("stddev"), 0.01);
assertEquals(2.875, skew.numericProperty("stddev"), 0.01);
assertEquals(0, uniform.numericProperty("skew"), 0.01);
assertEquals(0, uniformWithMissing.numericProperty("skew"), 0.01);
assertEquals(0, peak.numericProperty("skew"), 0.01);
assertEquals(1.86, skew.numericProperty("skew"), 0.01);
assertEquals(-1.557, uniform.numericProperty("kurtosis"), 0.01);
assertEquals(-1.557, uniformWithMissing.numericProperty("kurtosis"), 0.01);
assertEquals(0.5, peak.numericProperty("kurtosis"), 0.01);
assertEquals(1.983, skew.numericProperty("kurtosis"), 0.01);
}
@Test
public void testAutoConvertNumeric() {
Field a = Data.makeColumnField("field", null, new Object[] {1, 2, 3, 4});
Field b = Auto.convert(a);
assertEquals(a, b);
Assert.assertEquals(true, b.isNumeric());
Assert.assertEquals(false, b.isDate());
Assert.assertEquals(2.5, b.numericProperty("mean"), 0.001);
a = Data.makeColumnField("field", null, new Object[] {"1", "2", "3", "4"});
b = Auto.convert(a);
Assert.assertEquals(false, a == b);
Assert.assertEquals(true, b.isNumeric());
Assert.assertEquals(false, b.isDate());
Assert.assertEquals(2.5, b.numericProperty("mean"), 0.001);
a = Data.makeColumnField("field", null, new Object[] {"a", "2", "3", "4"});
b = Auto.convert(a);
Assert.assertEquals(false, a == b);
Assert.assertEquals(true, b.isNumeric());
Assert.assertEquals(false, b.isDate());
Assert.assertEquals(3.0, b.numericProperty("mean"), 0.001);
a = Data.makeColumnField("field", null, new Object[] {"a", "2", "c", "4"});
b = Auto.convert(a);
Assert.assertEquals(true, a == b);
Assert.assertEquals(false, b.isNumeric());
Assert.assertEquals(false, b.isDate());
}
