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;
}
@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());
}
@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 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 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 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);
}