public void Count_TwoItemsAdded_ReturnsTwo()
{
// arrange
string f1 = "F1";
string f2 = "F2";
FeatureNumericalManager featureManager = new FeatureNumericalManager();
featureManager.Add(new FeatureNumerical("F1"));
featureManager.Add(new FeatureNumerical("F2"));
ItemNumericalSet set = new ItemNumericalSet(featureManager);
ItemNumerical i1 = set.CreateItem();
i1.SetValue(f1, 3);
i1.SetValue(f2, 4.7f);
set.AddItem(i1);
ItemNumerical i2 = set.CreateItem();
i2.SetValue(f1, 1);
i2.SetValue(f2, 2.1f);
set.AddItem(i2);
// act
int qty = set.Count();
// assert
Assert.AreEqual(2, qty);
}
public void GetRandomSubset_9from15_Returns9()
{
// arrange
FillItemSet_Features4_Items15();
float subcountRatio = 0.6f;
// act
ItemNumericalSet subset = _set.GetRandomSubset(subcountRatio, false);
// assert
Assert.IsNotNull(subset);
Assert.AreEqual(9, subset.Count());
}
public void FillItemNumericalSet_ExcelDocument_()
{
// arrange
GenerateExcelDocumentF3();
FileInfo fi = new FileInfo(@"temp\test.xlsx");
//FileInfo fi = new FileInfo(@"data\test-4x1000.xlsx");
if (!fi.Exists)
{
Assert.Fail();
}
// act
ItemNumericalSet set = ExcelParser.ParseItemNumericalSet(fi.FullName, 1);
// assert
Assert.IsNotNull(set);
Assert.AreEqual(_itemsQty, set.Count());
Assert.AreEqual(true, set.CheckConsistency());
}
public FeatureNumericalSplitValue Split(ItemNumericalSet set, string resolutionFeatureName)
{
double totalIndex = set.GetRSS(resolutionFeatureName);
if (totalIndex == 0)
{
return(null);
}
double minIndex = totalIndex;
var featureNameList = set.GetFeatureNames();
var featureNames = featureNameList.Where(x => x != resolutionFeatureName);
if (!featureNames.Any())
{
throw new Exception();
}
FeatureNumericalSplitValue res = new FeatureNumericalSplitValue();
foreach (string fn in featureNames)
{
set.SortItems(fn);
int qty = set.Count();
for (int k = 1; k < qty; k++)
{
//ItemNumericalSet left = new ItemNumericalSet(featureNameList);
//ItemNumericalSet right = new ItemNumericalSet(featureNameList);
ItemNumericalSet left = set.Clone();
ItemNumericalSet right = set.Clone();
for (int i = 0; i < k; i++)
{
left.AddItem(set.GetItem(i));
}
for (int i = k; i < qty; i++)
{
right.AddItem(set.GetItem(i));
}
double leftIndex = left.GetRSS(resolutionFeatureName);
double rightIndex = right.GetRSS(resolutionFeatureName);
bool sumIndexIsLessThanMinIndex = leftIndex + rightIndex < minIndex;
bool sumIndexIsEqualToMinIndex = leftIndex + rightIndex == minIndex;
if (sumIndexIsLessThanMinIndex || (k == qty - 1 && sumIndexIsEqualToMinIndex))
{
if (sumIndexIsLessThanMinIndex)
{
minIndex = leftIndex + rightIndex;
}
res.FeatureName = fn;
double lv = left.GetItem(left.Count() - 1).GetValue(fn);
double rv = right.GetItem(0).GetValue(fn);
res.FeatureValue = Math.Round(lv + (rv - lv) / 2, 5);
res.Left = left;
res.Right = right;
}
}
}
return(res);
}
public FeatureNumericalSplitValue Split(ItemNumericalSet set, string resolutionFeatureName)
{
double totalIndex = set.GetGini2(resolutionFeatureName);
if (totalIndex == 0)
{
return(null);
}
double minIndex = totalIndex;
double maxDelta = 0;
int n = set.Count();
var featureNameList = set.GetFeatureNames();
var featureNames = featureNameList.Where(x => x != resolutionFeatureName);
if (!featureNames.Any())
{
throw new Exception();
}
FeatureNumericalSplitValue res = new FeatureNumericalSplitValue();
foreach (string fn in featureNames)
{
set.SortItems(fn);
int qty = set.Count();
for (int k = 1; k < qty; k++)
{
//ItemNumericalSet left = new ItemNumericalSet(featureNameList);
//ItemNumericalSet right = new ItemNumericalSet(featureNameList);
ItemNumericalSet left = set.Clone();
ItemNumericalSet right = set.Clone();
for (int i = 0; i < k; i++)
{
left.AddItem(set.GetItem(i));
}
for (int i = k; i < qty; i++)
{
right.AddItem(set.GetItem(i));
}
double leftIndex = left.GetGini2(resolutionFeatureName);
double rightIndex = right.GetGini2(resolutionFeatureName);
int lc = left.Count();
int rc = right.Count();
double delta = totalIndex - ((lc * 1.0) / n) * leftIndex - ((rc * 1.0) / n) * rightIndex;
if (delta > maxDelta)
{
maxDelta = delta;
res.FeatureName = fn;
double lv = left.GetItem(left.Count() - 1).GetValue(fn);
double rv = right.GetItem(0).GetValue(fn);
res.FeatureValue = Math.Round(lv + (rv - lv) / 2, 5);
res.Left = left;
res.Right = right;
}
}
}
return(res);
}
public int ItemCount()
{
return(_set.Count());
}