public static RowMatchObject FuzzyGENDERMatches(Row[] allData)
{
Console.WriteLine("Matching GENDER");
FastEditDistanceGrouper fastEditDistanceGrouper = new FastEditDistanceGrouper();
return(fastEditDistanceGrouper.DistanceAtMostN(allData, d => d.GENDER, 0));
}
public static RowMatchObject FuzzyADDRESS2Matches(Row[] allData)
{
Console.WriteLine("Matching ADDRESS2");
FastEditDistanceGrouper fastEditDistanceGrouper = new FastEditDistanceGrouper();
return(fastEditDistanceGrouper.DistanceAtMostN(allData, d => d.ADDRESS2, 2));
}
public static RowMatchObject FuzzyMIDDLEMatches(Row[] allData)
{
Console.WriteLine("Matching Middle Names");
FastEditDistanceGrouper fastEditDistanceGrouper = new FastEditDistanceGrouper();
return(fastEditDistanceGrouper.DistanceAtMostN(allData, d => d.MIDDLE, 2));
}
public static RowMatchObject FuzzySUFFIXMatches(Row[] allData)
{
Console.WriteLine("Matching SUFFIXES");
FastEditDistanceGrouper fastEditDistanceGrouper = new FastEditDistanceGrouper();
return(fastEditDistanceGrouper.DistanceAtMostN(allData, d => d.SUFFIX, 2));
}
public static RowMatchObject FuzzySSNMatches(Row[] allData)
{
Console.WriteLine("Matching SSN");
FastEditDistanceGrouper fastEditDistanceGrouper = new FastEditDistanceGrouper();
return(fastEditDistanceGrouper.DistanceAtMostN(allData, d => d.SSN <= 0 ? "" : d.SSN.ToString(), 2));
}
public static RowMatchObject FuzzyPhoneMatches(Row[] allData)
{
Console.WriteLine("Matching Phone");
FastEditDistanceGrouper fastGrouper = new FastEditDistanceGrouper();
return(fastGrouper.DistanceAtMostN(allData, d => d.PHONE <= 0 ? "" : d.PHONE.ToString(), 1));
}
public static RowMatchObject FuzzyFirstNameMatches(Row[] allData)
{
Console.WriteLine("Matching First Names");
FastEditDistanceGrouper fastEditDistanceGrouper = new FastEditDistanceGrouper();
return(fastEditDistanceGrouper.DistanceAtMostN(allData, d => d.FIRST, 2));
}
public static RowMatchObject FuzzyEMAILMatches(Row[] allData)
{
Console.WriteLine("Matching EMAIL");
FastEditDistanceGrouper fastEditDistanceGrouper = new FastEditDistanceGrouper();
return(fastEditDistanceGrouper.DistanceAtMostN(allData, d => d.EMAIL == "" ? "" : d.EMAIL.Split('@').First(), 2));
}
public static RowMatchObject FuzzyMOTHERS_MAIDEN_NAMEMatches(Row[] allData)
{
Console.WriteLine("Matching MOTHERS_MAIDEN_NAME");
FastEditDistanceGrouper fastEditDistanceGrouper = new FastEditDistanceGrouper();
return(fastEditDistanceGrouper.DistanceAtMostN(allData, d => d.MOTHERS_MAIDEN_NAME, 2));
}
public static RowMatchObject FuzzyZIPMatches(Row[] allData)
{
Console.WriteLine("Matching ZIP");
FastEditDistanceGrouper fastEditDistanceGrouper = new FastEditDistanceGrouper();
return(fastEditDistanceGrouper.DistanceAtMostN(allData, d => d.ZIP > 0 ? d.ZIP.ToString() : "", 2));
}
public static Matches FuzzyMatchOnNImportantFields(int[] eids, List <RowMatchObject> matchObjectsForFields, int n, Row[] allData)
{
int maxEid = eids.Max();
Matches toReturn = MatchesEngine.NewMatches(maxEid + 1);
FastEditDistanceGrouper fastEditDistanceGrouper = new FastEditDistanceGrouper();
int c = 0;
int[] eidToMatchCount = new int[maxEid + 1];
List <int> usedEids = new List <int>();
Random r = new Random();
foreach (int eid in eids)
{
usedEids.Clear();
Console.Write($"\r{c++}/{eids.Count()} Final Row Matches");
foreach (var matchObject in matchObjectsForFields)
{
int index = matchObject.EidToIndex[eid];
if (index == -1)
{
continue;
}
var neigborIndices = matchObject.Matches.Neighbors(index);
foreach (var neighborIndex in neigborIndices)
{
var neighborEids = matchObject.IndexToEids[neighborIndex.Index];
foreach (var neighborEid in neighborEids)
{
if (eid > neighborEid) //We will do the exact same computation when we find all the matches for neighborEID.
{
continue;
}
usedEids.Add(neighborEid);
eidToMatchCount[neighborEid]++;
}
}
}
int bigNeighborCount = 0;
foreach (var usedEid in usedEids)
{
if (eidToMatchCount[usedEid] >= n)
{
toReturn.AddMatch(eid, usedEid, eidToMatchCount[usedEid]); //One way of recording the NUMBER of fuzzy matches
bigNeighborCount++;
}
eidToMatchCount[usedEid] = 0;
}
//if (r.NextDouble() < .0001)
//{
// Console.WriteLine("Delete me");
// Console.WriteLine(bigNeighborCount);
// int betterCount = 0;
// var myGuy = allData.Where(d => d.EnterpriseID == eid).First();
// foreach (var datum in allData)
// {
// if (eid <= datum.EnterpriseID)
// {
// if (EasiestAgreementCount(myGuy, datum) >= n)
// {
// betterCount++;
// }
// }
// }
// Console.WriteLine(betterCount);
//}
}
Console.WriteLine("\nCleaning Two Field Fuzzy Match Object");
toReturn.Clean(); //I think I've actually staged things in a way that makes this unnecessary
Console.WriteLine("Done Cleaning");
return(toReturn);
}