public static void Gethistoricaldata(double[] testpointsx, List <double[]> userfeedbacks, out double usery)
{
usery = -1;
foreach (var fb in userfeedbacks)
{
var fbx = OnlineGPupdating.GetXfromTraining(OnlineGPupdating.RowtoMatrix(fb));
if (EvaluationForBinaryTree.IsSameImputs(testpointsx, fbx))
{
usery = fb[fb.Length - 1];
break;
}
else
{
usery = -1;
}
}
}
public static void WriteUserFeedback(double[] newdata, string csvaddress)
{
var fxnew = OnlineGPupdating.GetXfromTraining(newdata);
var fynew = newdata[newdata.Length - 1];
var olddata = ReadUserFeedback(csvaddress);
int deletindex = 0;
bool needtodelete = false;
for (int i = 0; i < olddata.Count; i++)
{
var f = olddata[i];
var fxold = OnlineGPupdating.GetXfromTraining(f);
if (EvaluationForBinaryTree.IsSameImputs(fxold, OnlineGPupdating.RowtoMatrix(fxnew)))
{
f[f.Length - 1] = fynew;
needtodelete = true;
deletindex = i;
break;
}
}
if (needtodelete)
{
olddata.RemoveAt(deletindex);
}
olddata.Add(newdata);
System.IO.File.Delete(csvaddress);
var csv = new StringBuilder();
foreach (var value in olddata)
{
var str = "";
for (int i = 0; i < value.Length; ++i)
{
str += value[i].ToString();
if (i < value.Length - 1)
{
str += ",";
}
}
csv.AppendLine(str);
}
// csv.AppendLine("Name,age");
File.AppendAllText(csvaddress, csv.ToString());
csv.Clear();
}
public static void Updatrainedtaclusters(double[] feedbackpointxy,
Dictionary <List <double[]>, List <double[, ]> > trainedclusters)
{
var existingpoint = false;
var deletindex = 0;
var mfeedbackpointx = GetXfromTraining(feedbackpointxy);
var mfeedbackpoint = RowtoMatrix(mfeedbackpointx);
var measures = new List <double>();
var preclosestcenter = Getpreclosestcenter(trainedclusters, mfeedbackpoint, out measures);
var maxmeasure = measures.Max();
var threahold = 1e-10;; //TBD
// var threahold = double.NegativeInfinity;
var closestclu = trainedclusters[trainedclusters.Keys.First(k => k[0].SequenceEqual(preclosestcenter))];
foreach (var point in closestclu)
{
var xpointm = GetXfromTraining(point);
var xfeedback = MatrixtoRow(GetXfromTraining(RowtoMatrix(feedbackpointxy)));
if (EvaluationForBinaryTree.IsSameImputs(xfeedback, xpointm))
{
existingpoint = true;
break;
}
else
{
deletindex++;
}
}
if (existingpoint)
{
var counter = 0;
var newclosestclu = closestclu;
for (int i = 0; i < feedbackpointxy.Length; i++)
{
newclosestclu[deletindex][0, i] = feedbackpointxy[i];
}
var samekey = trainedclusters.Keys.First(k => k[0].SequenceEqual(preclosestcenter));
trainedclusters.Remove(samekey);
trainedclusters.Add(samekey, newclosestclu);
}
if (maxmeasure > threahold) // add point to cluster
{
var newvalues = new List <double[, ]>(closestclu);
newvalues.Add(RowtoMatrix(feedbackpointxy));
if (newvalues.Count > 100)//remove data point
{
// var diaweight = GetdaigonalWeight(trainedclu.Keys.First()[2]);
// var farestcenterpoint = GetClosetPoint(closestclu, RowtoMatrix(feedbackpointxy), diaweight);
var oldkey = new List <double[]>(trainedclusters.Keys.First(k => k[0].SequenceEqual(preclosestcenter)));
var untrainedclu = new Dictionary <List <double[]>, List <double[, ]> >();
var removeindex = Gen.Random.Numbers.Integers(0, closestclu.Count - 1)();
newvalues.RemoveAt(removeindex);
var newcenter = new double[1, feedbackpointxy.Length];
foreach (var p in newvalues)
{
newcenter = newcenter.add(p);
}
newcenter = newcenter.divide(newvalues.Count);
oldkey[0] = MatrixtoRow(newcenter);
untrainedclu.Add(oldkey, newvalues);
var trainedclu = UpdataRegAndGP(untrainedclu);
trainedclusters.Remove(trainedclusters.Keys.First(k => k[0].SequenceEqual(preclosestcenter)));
trainedclusters.Add(trainedclu.Keys.First(), newvalues);
}
else//add point and update
{
var oldkey = new List <double[]>(trainedclusters.Keys.First(k => k[0].SequenceEqual(preclosestcenter)));
var untrainedclu = new Dictionary <List <double[]>, List <double[, ]> >();
var newcenter = new double[1, feedbackpointxy.Length];
foreach (var p in newvalues)
{
newcenter = newcenter.add(p);
}
newcenter = newcenter.divide(newvalues.Count);
oldkey[0] = MatrixtoRow(newcenter);
untrainedclu.Add(oldkey, newvalues);
var trainedclu = UpdataRegAndGP(untrainedclu);
trainedclusters.Remove(trainedclusters.Keys.First(k => k[0].SequenceEqual(preclosestcenter)));
trainedclusters.Add(trainedclu.Keys.First(), newvalues);
}
}
else //generate new cluster
{
var newcluter = Generatenewcluter(feedbackpointxy);
trainedclusters.Add(newcluter.Keys.First(), newcluter.Values.First());
}
}
public static void GetTimeAndSD(SubAssembly sub, double[] testpoints, string actionname, out double time,
out double SD)
{
time = 0;
SD = 0;
var inputdictionary = new Dictionary <List <double[]>, List <double[, ]> >();
double[,] subinputs;
double[] userx;
double usery, subtime, subsd;
var userfeedbacks = new List <double[]>();
if (actionname.StartsWith("S") || actionname.StartsWith("s"))
{
userfeedbacks = usersecuredata;
inputdictionary = SecureDictionary;
Gethistoricaldata(testpoints, userfeedbacks, out usery);
// usery = -1; // disable userfeedback
if (usery != -1)
{
time = usery;
SD = 0;
}
else
{
foreach (var otherf in sub.Secure.Fasteners)
{
if (otherf.SecureModelInputs != null)
{
if (EvaluationForBinaryTree.IsSameImputs(testpoints, otherf.SecureModelInputs) && otherf.Time > 0.001)
//if time is included in subassembly
{
time = otherf.Time;
SD = 0;
break;
}
else
{
TimeAndSD(testpoints, inputdictionary, userfeedbacks, out time, out SD);
break;
}
}
}
}
}
else
{
if (actionname.StartsWith("m") || actionname.StartsWith("M"))
{
userfeedbacks = usermovedata;
inputdictionary = MoveDictionary;
subinputs = sub.MoveRoateModelInputs;
subtime = sub.MoveRoate.Time;
subsd = sub.MoveRoate.TimeSD;
}
else if (actionname.StartsWith("i") || actionname.StartsWith("I"))
{
userfeedbacks = userinstalldata;
inputdictionary = InstallDictionary;
subinputs = sub.InstallModelInputs;
subtime = sub.Install.Time;
subsd = sub.Install.TimeSD;
}
else
{
userfeedbacks = userrotatedata;
inputdictionary = RotateDictionary;
subinputs = sub.RotateModelInputs;
subtime = sub.Rotate.Time;
subsd = sub.Rotate.TimeSD;
}
Gethistoricaldata(testpoints, userfeedbacks, out usery);
// usery = -1; // disable userfeedback
if (usery != -1)
{
time = usery;
SD = 0;
if (actionname.StartsWith("i") && time > 2.9 && time < 3.01)
{
var sss = 1;
}
}
else
{
if (EvaluationForBinaryTree.IsSameImputs(testpoints, subinputs) && subtime != 0)
//if time is included in subassembly
{
time = subtime;
SD = subsd;
if (actionname.StartsWith("i") && time > 2.9 && time < 3.01)
{
var sss = 1;
}
}
else
{
TimeAndSD(testpoints, inputdictionary, userfeedbacks, out time, out SD);
if (actionname.StartsWith("i") && time > 2.9 && time < 3.01)
{
var sss = 1;
}
}
}
}
}
