// TestStringTable
func TestStringTable(test *testing.T) {
test_strings := []string{"a", "b", "c", "d", "e"}
table := db.NewStringTable()
// Map every string twice. During the first cycle, the string
// should not be found. During the second cycle it should be
// found. Unmap should return the original value during both
// cycles.
for i := 0; i < 2; i++ {
for _, s := range test_strings {
m, ok := table.Encode(s)
u := table.Decode(m)
if u != s {
test.Errorf(`"%s" mapped back to "%s" during cycle %d`, s, u, i)
}
if ok && i == 0 {
test.Errorf(`"%s" found during fill cycle`, s)
}
if !ok && i == 1 {
test.Errorf(`"%s" not found during verify cycle`, s)
}
}
}
}
func featureFactory(s string) db.DataFrameFeature {
var feature db.Feature
switch s[0] {
case 'C':
feature = db.NewCategoricalFeature(db.NewStringTable())
default:
feature = db.NewNumericFeature(nil)
}
return db.NewDataFrameFeature(s, feature)
}
func TestCategoricalFeature(t *testing.T) {
// Assign to Feature to ensure CategoricalFeature implements Feature
cf := db.NewCategoricalFeature(db.NewStringTable())
testStrings := []string{"alpha", "beta", "delta", "beta", "alpha"}
for _, s := range testStrings {
cf.AddFromString(s)
}
check := func(index int, expected string) {
actual := cf.Value(index).(string)
altActual := cf.Decode(cf.NumericValue(index)).(string)
if actual != altActual {
t.Errorf("Value(%d) returned %v; Decode(NumericValue(%d) returned %v", index, actual, index, altActual)
}
if actual != expected {
t.Errorf("Get(%d) got %v; expecting %v", index, actual, expected)
}
}
for i, s := range testStrings {
check(i, s)
}
if cf.Len() != len(testStrings) {
t.Errorf("Length() returned %d; expecting %d", cf.Len(), 5)
}
ordercheck := func(index int, expected string) {
orderedString := cf.Decode(cf.NumericValue(cf.InOrder(index)))
if orderedString != expected {
t.Errorf("Get(%d) got %s; expecting %s", index, orderedString, expected)
}
}
cf.Prepare()
ordercheck(0, "alpha")
ordercheck(1, "alpha")
ordercheck(2, "beta")
ordercheck(3, "beta")
ordercheck(4, "delta")
if n := cf.Categories(); n != 3 {
t.Errorf("Expected %d categories; got %d\n", 3, n)
}
}
