public FPNode(string item, int support, FPNode parent)
{
_item = item;
_support = support;
_parent = parent;
_children = new List<FPNode>();
}
public ItemHeaderElement(string itemID, int support, FPNode nodeLink)
{
_nodeLinks = new List <FPNode>();
_nodeLinks.Add(nodeLink);
_itemID = itemID;
_support = support;
}
public FPNode(string item, int support, FPNode parent)
{
_item = item;
_support = support;
_parent = parent;
_children = new List <FPNode>();
}
/* add the items in dto to tree recursively && add node links to itemHeaderTable */
private void addToTree(ref FPNode root, TransactionDto dto, ref List <ItemHeaderElement> itemHeaderTable)
{
if (dto.items.Count == 0)
{
return;
}
FPNode parent = root;
foreach (FPNode node in root.children)
{
if (node.item == dto.items[0])
{
parent = node;
}
}
if (parent == root)
{
FPNode fpn = new FPNode(dto.items[0], 1, parent);
parent.children.Add(fpn);
parent = fpn;
for (int i = 0; i < itemHeaderTable.Count; i++)
{
if (itemHeaderTable[i].itemID == dto.items[0])
{
itemHeaderTable[i].addNodeLink(fpn);
}
}
}
else
{
parent.support = parent.support + 1;
}
dto.items.RemoveAt(0);
addToTree(ref parent, dto, ref itemHeaderTable);
}
private void mineConditionalFPTree(ref List <ItemHeaderElement> conditionalItemHeader, ref List <FrequentPattern> frequentPatterns, string suffix)
{
foreach (ItemHeaderElement ihe in conditionalItemHeader)
{
List <string> items = new List <string>();
items.Add(suffix);
items.Add(ihe.itemID);
items.Reverse();
FrequentPattern fp = new FrequentPattern(items, ihe.support);
frequentPatterns.Add(fp);
}
foreach (ItemHeaderElement ihe in conditionalItemHeader)
{
foreach (FPNode fpn in ihe.nodeLinks)
{
List <string> items = new List <string>();
FPNode aFpn = fpn;
int support = aFpn.support;
while (aFpn.item != null)
{
support = min(support, aFpn.support);
items.Add(aFpn.item);
aFpn = aFpn.parent;
}
if (items.Count > 1 && support >= min_sup)
{
items.Reverse();
items.Add(suffix);
FrequentPattern fp = new FrequentPattern(items, support);
frequentPatterns.Add(fp);
}
}
}
}
public List <FrequentPattern> mine()
{
var item = new List <string>();
var support = new List <int>();
int index;
/* get support count of each item */
foreach (TransactionDto dto in dtos)
{
foreach (string p in dto.items)
{
index = item.IndexOf(p);
if (index == -1) //not found
{
item.Add(p);
support.Add(1);
}
else
{
support[index] = support[index] + 1;
}
} //foreach
} //foreach
/* sort in desc order based on support...thank you .NET 4.0!! */
var orderedSupport = support.Zip(item, (x, y) => new { x, y }).OrderBy(pair => pair.x).ToList();
support = orderedSupport.Select(pair => pair.x).ToList();
item = orderedSupport.Select(pair => pair.y).ToList();
support.Reverse();
item.Reverse();
/* we are now sorted */
/*
* Console.WriteLine("Frequent 1-itemsets");
* for (int i = 0; i < support.Count; i++)
* {
* Console.WriteLine(item[i] + " --> " + support[i]);
* }
* Console.WriteLine("Press any key to continue...");
* Console.ReadKey();
*/
/* create item header table */
List <ItemHeaderElement> itemHeaderTable = new List <ItemHeaderElement>();
for (int i = 0; i < support.Count; i++)
{
if (support[i] >= min_sup)
{
itemHeaderTable.Add(new ItemHeaderElement(item[i], support[i]));
if (support[i] >= min_sup)
{
frequentPatterns.Add(new FrequentPattern(item[i], support[i])); //CAN: Add singleton sets to frequent pattern list
}
}
}
/*
* Console.WriteLine("Create ItemHeaderTable");
* Console.WriteLine("ItemID | Support Count | Node-Link");
* for (int i = 0; i < item.Count; i++) Console.WriteLine(item[i] + " | " + support[i] + " | NULL");
* Console.WriteLine("Press any key to continue...");
* Console.ReadKey();
*/
List <string> items;
List <int> lsupport;
/* sort all transaction items in L order */
foreach (TransactionDto dto in dtos)
{
items = dto.items;
lsupport = new List <int>();
foreach (string i in items)
{
lsupport.Add(support[item.IndexOf(i)]);
}
orderedSupport = lsupport.Zip(items, (x, y) => new { x, y }).OrderBy(pair => pair.y).Reverse().OrderBy(pair => pair.x).ToList();
items = orderedSupport.Select(pair => pair.y).ToList();
items.Reverse();
dto.items = items;
}
/*
* Console.WriteLine("Tranactions sorted in L order");
* for (int i = 0; i < dtos.Count; i++)
* {
* string strItem = string.Empty;
* foreach (string s in dtos[i].items)
* {
* strItem += s + ", ";
* }
* strItem = strItem.Substring(0, strItem.Length - 2);
* Console.WriteLine(dtos[i].tid + "| " + strItem);
* }
* Console.WriteLine("Press any key to continue...");
* Console.ReadKey();
*/
/* start building FP-Tree */
FPNode root = new FPNode(null, 0);
foreach (TransactionDto dto in dtos)
{
addToTree(ref root, dto, ref itemHeaderTable);
}
/*
* Console.WriteLine("ItemHeaderTable w/node-links");
*
* foreach (ItemHeaderElement e in itemHeaderTable)
* {
* Console.WriteLine(e.ToString());
* }
* Console.WriteLine("Press any key to continue...");
* Console.ReadKey();
*
* Console.WriteLine("Root branches");
*
* foreach (FPNode aNode in root.children)
* {
* Console.WriteLine("----");
* FPNode myNode = aNode;
* while (myNode != null)
* {
* Console.WriteLine(myNode.item + " --> ");
* if (myNode.children.Count == 0) myNode = null;
* else myNode = myNode.children[0];
* }
* }
* Console.WriteLine("Press any key to continue...");
* Console.ReadKey();
*/
/* Finally, let's mine the tree! */
/* start at the bottom because we want to work our way up the tree from the leafs */
//for simplicity I will reverse the list
itemHeaderTable.Reverse();
/*
* Console.WriteLine("Reversed ItemHeaderTable w/node-links");
* foreach (ItemHeaderElement e in itemHeaderTable)
* {
* Console.WriteLine(e.ToString());
* }
* Console.WriteLine("Press any key to continue...");
* Console.ReadKey();
*/
mineTheTree(ref itemHeaderTable, ref frequentPatterns);
return(frequentPatterns.OrderBy(x => x.support).ThenBy(x => x.items.Count).Reverse().ToList());
}//mine()
private void createConditionalFPTree(List <FPPrefixPath> prefixPaths, ref List <ItemHeaderElement> itemHeaders)
{
foreach (FPPrefixPath p in prefixPaths)
{
for (int i = 0; i < p.support; i++)
{
for (int j = 0; j < p.prefixpath.Count; j++)
{
Boolean found = false;
for (int k = 0; k < itemHeaders.Count; k++)
{
if (itemHeaders[k].itemID == p.prefixpath[j].item)
{
found = true;
itemHeaders[k].support = itemHeaders[k].support + 1;
}
}
if (!found)
{
itemHeaders.Add(new ItemHeaderElement(p.prefixpath[j].item, 1));
}
}
}
}
/* remove the itemHeaders that don't meet min_sup */
for (int n = 0; n < itemHeaders.Count; n++)
{
if (itemHeaders[n].support < min_sup)
{
itemHeaders.RemoveAt(n--);
}
}
/*
* Console.WriteLine("ConditionalHeaderTable wo/node-links");
* foreach (ItemHeaderElement e in itemHeaders)
* {
* Console.WriteLine(e.ToString());
* }
* Console.WriteLine("Press any key to continue...");
* Console.ReadKey();
*/
/* remove prefixpath nodes that don't meet min_sup */
foreach (FPPrefixPath p in prefixPaths)
{
List <int> removeList = new List <int>();
p.prefixpath.Reverse();
for (int i = 0; i < p.prefixpath.Count; i++)
{
bool found = false;
for (int j = 0; j < itemHeaders.Count; j++)
{
if (itemHeaders[j].itemID == p.prefixpath[i].item)
{
found = true;
}
}
if (!found)
{
removeList.Add(i);
}
}
for (int i = 0; i < removeList.Count; i++)
{
p.prefixpath.RemoveAt(removeList[i] - i); //CAN: Added because indexes change after the removing an element. Did not encounter this issue in testing
}
}
FPNode root = new FPNode(null, 0);
foreach (FPPrefixPath p in prefixPaths)
{
TransactionDto dto = new TransactionDto();
foreach (FPNode fpn in p.prefixpath)
{
dto.items.Add(fpn.item);
}
/*
* Console.WriteLine("Adding to the tree " + p.support + " times: ");
* Console.WriteLine(dto);
* Console.ReadKey();
*/
for (int i = 0; i < p.support; i++)
{
TransactionDto tmpDto = new TransactionDto();
foreach (string item in dto.items)
{
tmpDto.items.Add(item);
}
addToTree(ref root, tmpDto, ref itemHeaders);
}
}
/*
* Console.WriteLine("ConditionalHeaderTable w/node-links");
* foreach (ItemHeaderElement e in itemHeaders)
* {
* Console.WriteLine(e.ToString());
* }
* Console.WriteLine("Press any key to continue...");
* Console.ReadKey();
*
* Console.WriteLine("Conditional FP-Tree children");
* foreach (FPNode aNode in root.children)
* {
* Console.WriteLine("----");
* FPNode myNode = aNode;
* while (myNode != null)
* {
* Console.WriteLine(myNode.item + ":" + myNode.support + " --> ");
* if (myNode.children.Count == 0) myNode = null;
* else myNode = myNode.children[0];
* }
* }
* Console.WriteLine("Press any key to continue...");
* Console.ReadKey();
*/
}
public void addNodeLink(FPNode node)
{
_nodeLinks.Add(node);
}
public FPPrefixPath(List <FPNode> prefixpath, FPNode suffix)
{
_prefixpath = prefixpath;
_suffix = suffix;
_support = suffix.support;
}
public List<FrequentPattern> mine()
{
var item = new List<string>();
var support = new List<int>();
int index;
/* get support count of each item */
foreach (TransactionDto dto in dtos)
{
foreach (string p in dto.items)
{
index = item.IndexOf(p);
if (index == -1) //not found
{
item.Add(p);
support.Add(1);
}
else
{
support[index] = support[index] + 1;
}
}//foreach
}//foreach
/* sort in desc order based on support...thank you .NET 4.0!! */
var orderedSupport = support.Zip(item, (x, y) => new { x, y }).OrderBy(pair => pair.x).ToList();
support = orderedSupport.Select(pair => pair.x).ToList();
item = orderedSupport.Select(pair => pair.y).ToList();
support.Reverse();
item.Reverse();
/* we are now sorted */
/*
Console.WriteLine("Frequent 1-itemsets");
for (int i = 0; i < support.Count; i++)
{
Console.WriteLine(item[i] + " --> " + support[i]);
}
Console.WriteLine("Press any key to continue...");
Console.ReadKey();
*/
/* create item header table */
List<ItemHeaderElement> itemHeaderTable = new List<ItemHeaderElement>();
for (int i = 0; i < support.Count; i++)
{
if (support[i] >= min_sup)
{
itemHeaderTable.Add(new ItemHeaderElement(item[i], support[i]));
if (support[i] >= min_sup) frequentPatterns.Add(new FrequentPattern(item[i], support[i])); //CAN: Add singleton sets to frequent pattern list
}
}
/*
Console.WriteLine("Create ItemHeaderTable");
Console.WriteLine("ItemID | Support Count | Node-Link");
for (int i = 0; i < item.Count; i++) Console.WriteLine(item[i] + " | " + support[i] + " | NULL");
Console.WriteLine("Press any key to continue...");
Console.ReadKey();
*/
List<string> items;
List<int> lsupport;
/* sort all transaction items in L order */
foreach (TransactionDto dto in dtos)
{
items = dto.items;
lsupport = new List<int>();
foreach (string i in items)
{
lsupport.Add(support[item.IndexOf(i)]);
}
orderedSupport = lsupport.Zip(items, (x, y) => new { x, y }).OrderBy(pair => pair.y).Reverse().OrderBy(pair => pair.x).ToList();
items = orderedSupport.Select(pair => pair.y).ToList();
items.Reverse();
dto.items = items;
}
/*
Console.WriteLine("Tranactions sorted in L order");
for (int i = 0; i < dtos.Count; i++)
{
string strItem = string.Empty;
foreach (string s in dtos[i].items)
{
strItem += s + ", ";
}
strItem = strItem.Substring(0, strItem.Length - 2);
Console.WriteLine(dtos[i].tid + "| " + strItem);
}
Console.WriteLine("Press any key to continue...");
Console.ReadKey();
*/
/* start building FP-Tree */
FPNode root = new FPNode(null, 0);
foreach (TransactionDto dto in dtos)
{
addToTree(ref root, dto, ref itemHeaderTable);
}
/*
Console.WriteLine("ItemHeaderTable w/node-links");
foreach (ItemHeaderElement e in itemHeaderTable)
{
Console.WriteLine(e.ToString());
}
Console.WriteLine("Press any key to continue...");
Console.ReadKey();
Console.WriteLine("Root branches");
foreach (FPNode aNode in root.children)
{
Console.WriteLine("----");
FPNode myNode = aNode;
while (myNode != null)
{
Console.WriteLine(myNode.item + " --> ");
if (myNode.children.Count == 0) myNode = null;
else myNode = myNode.children[0];
}
}
Console.WriteLine("Press any key to continue...");
Console.ReadKey();
*/
/* Finally, let's mine the tree! */
/* start at the bottom because we want to work our way up the tree from the leafs */
//for simplicity I will reverse the list
itemHeaderTable.Reverse();
/*
Console.WriteLine("Reversed ItemHeaderTable w/node-links");
foreach (ItemHeaderElement e in itemHeaderTable)
{
Console.WriteLine(e.ToString());
}
Console.WriteLine("Press any key to continue...");
Console.ReadKey();
*/
mineTheTree(ref itemHeaderTable, ref frequentPatterns);
return frequentPatterns.OrderBy(x => x.support).ThenBy(x => x.items.Count).Reverse().ToList();
}
private void createConditionalFPTree(List<FPPrefixPath> prefixPaths, ref List<ItemHeaderElement> itemHeaders)
{
foreach (FPPrefixPath p in prefixPaths)
{
for (int i = 0; i < p.support; i++)
{
for (int j = 0; j < p.prefixpath.Count; j++)
{
Boolean found = false;
for (int k=0; k<itemHeaders.Count; k++){
if (itemHeaders[k].itemID == p.prefixpath[j].item) {
found = true;
itemHeaders[k].support = itemHeaders[k].support + 1;
}
}
if (!found) itemHeaders.Add(new ItemHeaderElement(p.prefixpath[j].item, 1));
}
}
}
/* remove the itemHeaders that don't meet min_sup */
for (int n = 0; n < itemHeaders.Count; n++)
{
if (itemHeaders[n].support < min_sup)
{
itemHeaders.RemoveAt(n--);
}
}
/*
Console.WriteLine("ConditionalHeaderTable wo/node-links");
foreach (ItemHeaderElement e in itemHeaders)
{
Console.WriteLine(e.ToString());
}
Console.WriteLine("Press any key to continue...");
Console.ReadKey();
*/
/* remove prefixpath nodes that don't meet min_sup */
foreach (FPPrefixPath p in prefixPaths)
{
List<int> removeList = new List<int>();
p.prefixpath.Reverse();
for (int i = 0; i < p.prefixpath.Count; i++)
{
bool found = false;
for (int j = 0; j < itemHeaders.Count; j++)
{
if (itemHeaders[j].itemID == p.prefixpath[i].item)
{
found = true;
}
}
if (!found) removeList.Add(i);
}
for (int i = 0; i < removeList.Count; i++)
{
p.prefixpath.RemoveAt(removeList[i] - i); //CAN: Added because indexes change after the removing an element. Did not encounter this issue in testing
}
}
FPNode root = new FPNode(null, 0);
foreach (FPPrefixPath p in prefixPaths)
{
TransactionDto dto = new TransactionDto();
foreach (FPNode fpn in p.prefixpath)
{
dto.items.Add(fpn.item);
}
/*
Console.WriteLine("Adding to the tree " + p.support + " times: ");
Console.WriteLine(dto);
Console.ReadKey();
*/
for (int i = 0; i < p.support; i++)
{
TransactionDto tmpDto = new TransactionDto();
foreach (string item in dto.items)
{
tmpDto.items.Add(item);
}
addToTree(ref root, tmpDto, ref itemHeaders);
}
}
/*
Console.WriteLine("ConditionalHeaderTable w/node-links");
foreach (ItemHeaderElement e in itemHeaders)
{
Console.WriteLine(e.ToString());
}
Console.WriteLine("Press any key to continue...");
Console.ReadKey();
Console.WriteLine("Conditional FP-Tree children");
foreach (FPNode aNode in root.children)
{
Console.WriteLine("----");
FPNode myNode = aNode;
while (myNode != null)
{
Console.WriteLine(myNode.item + ":" + myNode.support + " --> ");
if (myNode.children.Count == 0) myNode = null;
else myNode = myNode.children[0];
}
}
Console.WriteLine("Press any key to continue...");
Console.ReadKey();
*/
}
/* add the items in dto to tree recursively && add node links to itemHeaderTable */
private void addToTree(ref FPNode root, TransactionDto dto, ref List<ItemHeaderElement> itemHeaderTable)
{
if (dto.items.Count == 0) return;
FPNode parent = root;
foreach (FPNode node in root.children)
{
if (node.item == dto.items[0])
{
parent = node;
}
}
if (parent == root)
{
FPNode fpn = new FPNode(dto.items[0], 1, parent);
parent.children.Add(fpn);
parent = fpn;
for (int i = 0; i < itemHeaderTable.Count; i++)
{
if (itemHeaderTable[i].itemID == dto.items[0]) itemHeaderTable[i].addNodeLink(fpn);
}
}
else
{
parent.support = parent.support + 1;
}
dto.items.RemoveAt(0);
addToTree(ref parent, dto, ref itemHeaderTable);
}