### Re: [Flashcoders] Determining probablity of a random pick.

Jiri wrote: Hello list, I have an XMLList that can vary in length. I pick a random node each time using the simple method below: code var tRandom:int = int(Math.random()*tXMLSource.nodes.length() ); return tXMLSource.nodes[tRandom] /code Now i would like to change it, so that I can specify a weight to each node. This weight would have to determine the probability the nodes get selected out when a random pick is being performed. Does anybody know how to build such an algorithm? Sum all of the node weights, then generate a random number within that range. Now you have to find the node corresponding to that number. This can be done by traversing the nodes in sequence. As you traverse the nodes sum the weight until the value generated is within the range - then you have the right node. You can optimise this by using an array of pointers, or keeping the weights in an external array, but that may be more or less interesting depending on the number of nodes involved and what optimisation you need. Paul Much appreciated, Jiri ___ Flashcoders mailing list Flashcoders@chattyfig.figleaf.com http://chattyfig.figleaf.com/mailman/listinfo/flashcoders ___ Flashcoders mailing list Flashcoders@chattyfig.figleaf.com http://chattyfig.figleaf.com/mailman/listinfo/flashcoders

### Re: [Flashcoders] Determining probablity of a random pick.

Thank you i understand it. Could u please elloborate on the last part ..You can optimise this by using an array of pointers.. Do you mean that this; a weight=5 b weight=1 c weight=2 First calculated the range for each node. A = 0 - 5 B = 6 - 7 C = 8 - 10 Then store in an array like so? Array[0-5] = A Array[6-7] = B Array[8-10] = C Last step get a random number between 0 - 10 = for instance 5 //result is then return Array[5] as XML If this is what you mean, would this approach not take a 'relatively' big amount of memory? JIri Paul Andrews wrote: Jiri wrote: Hello list, I have an XMLList that can vary in length. I pick a random node each time using the simple method below: code var tRandom:int = int(Math.random()*tXMLSource.nodes.length() ); return tXMLSource.nodes[tRandom] /code Now i would like to change it, so that I can specify a weight to each node. This weight would have to determine the probability the nodes get selected out when a random pick is being performed. Does anybody know how to build such an algorithm? Sum all of the node weights, then generate a random number within that range. Now you have to find the node corresponding to that number. This can be done by traversing the nodes in sequence. As you traverse the nodes sum the weight until the value generated is within the range - then you have the right node. You can optimise this by using an array of pointers, or keeping the weights in an external array, but that may be more or less interesting depending on the number of nodes involved and what optimisation you need. Paul Much appreciated, Jiri ___ Flashcoders mailing list Flashcoders@chattyfig.figleaf.com http://chattyfig.figleaf.com/mailman/listinfo/flashcoders ___ Flashcoders mailing list Flashcoders@chattyfig.figleaf.com http://chattyfig.figleaf.com/mailman/listinfo/flashcoders ___ Flashcoders mailing list Flashcoders@chattyfig.figleaf.com http://chattyfig.figleaf.com/mailman/listinfo/flashcoders

### Re: [Flashcoders] Determining probablity of a random pick.

Jiri wrote: Thank you i understand it. Could u please elloborate on the last part ..You can optimise this by using an array of pointers.. Do you mean that this; a weight=5 b weight=1 c weight=2 First calculated the range for each node. A = 0 - 5 B = 6 - 7 C = 8 - 10 Then store in an array like so? Array[0-5] = A Array[6-7] = B Array[8-10] = C Last step get a random number between 0 - 10 = for instance 5 //result is then return Array[5] as XML If this is what you mean, would this approach not take a 'relatively' big amount of memory? You don't have to store the node itself, just a reference to it, so no it wouldn't be a memory hog. It could even just be the numerical index of the node. Paul JIri Paul Andrews wrote: Jiri wrote: Hello list, I have an XMLList that can vary in length. I pick a random node each time using the simple method below: code var tRandom:int = int(Math.random()*tXMLSource.nodes.length() ); return tXMLSource.nodes[tRandom] /code Now i would like to change it, so that I can specify a weight to each node. This weight would have to determine the probability the nodes get selected out when a random pick is being performed. Does anybody know how to build such an algorithm? Sum all of the node weights, then generate a random number within that range. Now you have to find the node corresponding to that number. This can be done by traversing the nodes in sequence. As you traverse the nodes sum the weight until the value generated is within the range - then you have the right node. You can optimise this by using an array of pointers, or keeping the weights in an external array, but that may be more or less interesting depending on the number of nodes involved and what optimisation you need. Paul Much appreciated, Jiri ___ Flashcoders mailing list Flashcoders@chattyfig.figleaf.com http://chattyfig.figleaf.com/mailman/listinfo/flashcoders ___ Flashcoders mailing list Flashcoders@chattyfig.figleaf.com http://chattyfig.figleaf.com/mailman/listinfo/flashcoders ___ Flashcoders mailing list Flashcoders@chattyfig.figleaf.com http://chattyfig.figleaf.com/mailman/listinfo/flashcoders ___ Flashcoders mailing list Flashcoders@chattyfig.figleaf.com http://chattyfig.figleaf.com/mailman/listinfo/flashcoders

### Re: [Flashcoders] Determining probablity of a random pick.

An easy way is to if say you have 3 objects in a list given a weight options[0].weight = 1; options[1].weight = 3; options[2].weight = 4; instead of having it have three selections, you could add the weights and make it have 8 to choose from that point to the option. choice[0]=0; choice[1]=1; choice[2]=1; choice[3]=1; choice[4]=2; choice[5]=2; choice[6]=2; choice[7]=2; Jiri wrote: Hello list, I have an XMLList that can vary in length. I pick a random node each time using the simple method below: code var tRandom:int = int(Math.random()*tXMLSource.nodes.length() ); return tXMLSource.nodes[tRandom] /code Now i would like to change it, so that I can specify a weight to each node. This weight would have to determine the probability the nodes get selected out when a random pick is being performed. Does anybody know how to build such an algorithm? Much appreciated, Jiri ___ Flashcoders mailing list Flashcoders@chattyfig.figleaf.com http://chattyfig.figleaf.com/mailman/listinfo/flashcoders ___ Flashcoders mailing list Flashcoders@chattyfig.figleaf.com http://chattyfig.figleaf.com/mailman/listinfo/flashcoders

### Re: [Flashcoders] Determining probablity of a random pick.

I see, that makes sense. Cheers. Jiri Paul Andrews wrote: Jiri wrote: Thank you i understand it. Could u please elloborate on the last part ..You can optimise this by using an array of pointers.. Do you mean that this; a weight=5 b weight=1 c weight=2 First calculated the range for each node. A = 0 - 5 B = 6 - 7 C = 8 - 10 Then store in an array like so? Array[0-5] = A Array[6-7] = B Array[8-10] = C Last step get a random number between 0 - 10 = for instance 5 //result is then return Array[5] as XML If this is what you mean, would this approach not take a 'relatively' big amount of memory? You don't have to store the node itself, just a reference to it, so no it wouldn't be a memory hog. It could even just be the numerical index of the node. Paul JIri Paul Andrews wrote: Jiri wrote: Hello list, I have an XMLList that can vary in length. I pick a random node each time using the simple method below: code var tRandom:int = int(Math.random()*tXMLSource.nodes.length() ); return tXMLSource.nodes[tRandom] /code Now i would like to change it, so that I can specify a weight to each node. This weight would have to determine the probability the nodes get selected out when a random pick is being performed. Does anybody know how to build such an algorithm? Sum all of the node weights, then generate a random number within that range. Now you have to find the node corresponding to that number. This can be done by traversing the nodes in sequence. As you traverse the nodes sum the weight until the value generated is within the range - then you have the right node. You can optimise this by using an array of pointers, or keeping the weights in an external array, but that may be more or less interesting depending on the number of nodes involved and what optimisation you need. Paul Much appreciated, Jiri ___ Flashcoders mailing list Flashcoders@chattyfig.figleaf.com http://chattyfig.figleaf.com/mailman/listinfo/flashcoders ___ Flashcoders mailing list Flashcoders@chattyfig.figleaf.com http://chattyfig.figleaf.com/mailman/listinfo/flashcoders ___ Flashcoders mailing list Flashcoders@chattyfig.figleaf.com http://chattyfig.figleaf.com/mailman/listinfo/flashcoders ___ Flashcoders mailing list Flashcoders@chattyfig.figleaf.com http://chattyfig.figleaf.com/mailman/listinfo/flashcoders ___ Flashcoders mailing list Flashcoders@chattyfig.figleaf.com http://chattyfig.figleaf.com/mailman/listinfo/flashcoders