|
 |
| | | Author: | bb414989  | | Posted: | Aug 16, 2013 04:25 | | Subject: | Improve Wanted List / Shop By | | Viewed: | 108 times | | Topic: | Suggestions | | Status: | Implemented | |
|
| I'm new to Bricklink and I had an idea to improve the "Wanted List / Shop
By."
Is it possible to program a purchase suggestion?
Let me explain (for example):
- I have 20 different items in my Wanted List
- I can not have everything at the same vendor, there will be x minium sellers
- I'd like BrickLink suggest me the best way to buy at the cheapest my objects
from these sellers (eg 5 items the seller "A", 10 the seller "B" and 5 to the
seller "C")
Do you think it is a good idea?
|
|
| | | | | |
| | | | | Author: | RobErNat | | Posted: | Aug 16, 2013 04:50 | | Subject: | Re: Improve Wanted List / Shop By | | Viewed: | 40 times | | Topic: | Suggestions | |
|
| In Suggestions, Magikbibi writes:
| | I'm new to Bricklink and I had an idea to improve the "Wanted List / Shop
By."
Is it possible to program a purchase suggestion?
Let me explain (for example):
- I have 20 different items in my Wanted List
- I can not have everything at the same vendor, there will be x minium sellers
- I'd like BrickLink suggest me the best way to buy at the cheapest my objects
from these sellers (eg 5 items the seller "A", 10 the seller "B" and 5 to the
seller "C")
Do you think it is a good idea?
|
Well the idea sounds good, but I suspect it would require too many system resources.
In 3 days 6000 orders have been processed on Bricklink, so imagine how many times
per day people fill their carts, check or don't check out. If on top the
system needs to 'calculate' on wanted lists, well...
And on top the system cannot take in consideration shippingcosts and or other
costs, so the calculation would be false if based on items value only.
Sometimes it is cheaper to buy more expensive stuff and have less shipping if
you buy close by compared to another seller who has the items cheaper, but higher
expensive shipping.
Soo many many parameters 
|
|
|
| | | | | | | | | |
| | | | | | | Author: | bb414989 | | Posted: | Aug 16, 2013 05:15 | | Subject: | Re: Improve Wanted List / Shop By | | Viewed: | 32 times | | Topic: | Suggestions | |
|
| Yes indeed the request would be complex to implement. But I think at first, when
you click on "Wanted List / Shop By", there could be the total amount shown live
before entering the store.
While sometimes a cheaper room costs more expensive because of the delivery.
But personally when you filter in the country, postal rates are identical (as
regulated) ... and therefore would not be a valeure to consider.
|
|
| | | | | | | | | | | | | |
| | | | | | | | | Author: | mathematical | | Posted: | Aug 16, 2013 08:31 | | Subject: | Re: Improve Wanted List / Shop By | | Viewed: | 29 times | | Topic: | Suggestions | |
|
| Well...
If the program also factored in minimum buy (to reduce unneeded parts bought)
and shipping charges, the problem is actually a version of the traveling salesman
problem and is NP-hard, (Non-deterministic Polynomial-time hard) (You
can look both of those up on Wikipedia or something!). So, it is possible
for an implementation, but the time to solution cannot be solved in polynomial
time (read: reasonable time), unless P=NP (which has not been shown to be true,
but neither has P~=NP).
I don't mean to be mean or to burst your bubble or anything, but there's
just no way Bricklink would have enough computing power to accomplish this.
|
|
| | | | | | | | | | | | | | | | | |
| | | | | | | | | | | Author: | Timothy_Smith | | Posted: | Aug 16, 2013 08:51 | | Subject: | Re: Improve Wanted List / Shop By | | Viewed: | 24 times | | Topic: | Suggestions | |
|
| In Suggestions, mathematical writes:
| | Well...
If the program also factored in minimum buy (to reduce unneeded parts bought)
and shipping charges, the problem is actually a version of the traveling salesman
problem and is NP-hard, (Non-deterministic Polynomial-time hard) (You
can look both of those up on Wikipedia or something!). So, it is possible
for an implementation, but the time to solution cannot be solved in polynomial
time (read: reasonable time), unless P=NP (which has not been shown to be true,
but neither has P~=NP).
I don't mean to be mean or to burst your bubble or anything, but there's
just no way Bricklink would have enough computing power to accomplish this.
|
As a general problem yes- but for the specifically limited case of real wanted
lists against BL's actual inventory, it might be feasible using massively
parallel computing. Run it a distributed app across the network of folks who
have signed up for the service, like SETI@Home.
|
|
|
| | | | | | | | | | | | | | | | | |
| | | | | | | | | | | Author: | mhn1957 | | Posted: | Aug 16, 2013 09:00 | | Subject: | Re: Improve Wanted List / Shop By | | Viewed: | 25 times | | Topic: | Suggestions | |
|
| In Suggestions, mathematical writes:
| | Well...
If the program also factored in minimum buy (to reduce unneeded parts bought)
and shipping charges, the problem is actually a version of the traveling salesman
problem and is NP-hard, (Non-deterministic Polynomial-time hard) (You
can look both of those up on Wikipedia or something!). So, it is possible
for an implementation, but the time to solution cannot be solved in polynomial
time (read: reasonable time), unless P=NP (which has not been shown to be true,
but neither has P~=NP).
I don't mean to be mean or to burst your bubble or anything, but there's
just no way Bricklink would have enough computing power to accomplish this.
|
I think we could get closer to a solution by changing the way lots are matched,
or possibly adding a sequence/level of importance to the wanted lists or items
in them. To give an example, I added zamor spheres to my wanted list and that
one item destroyed any hope of getting a meaningful list of stores until I adjusted
the price down to $.03. On my wanted list, I set the price of things that are
secondary very low, or set the quantity very high, so that, after I find a store
that has most of the top tier items, I can press the (all items) button and get
a more comprehensive list. That's only after I've selected a couple of
stores to check. The wanted list maintenance is a little clunky, and I can think
of a number of ways using a 2 or 3 level search would work without attempting
a cpu intensive search which would bring up multiple stores as in the suggestion.
If I could specify that all zamor spheres should be considered one lot as far
as my search is concerned, that would help. The total number would be more important
to me than what colors, as I don't care about color in this instance. Since
they are listed by color, and some stores have multiple lots of a color, including
this item really distorts things.
In natural language,
Get a list of all stores that have the all of my level one items. (Required items.)
Take that list and display the stores, ranking them by the lot count and quantity
of level two items, based on my sort preference. Lot count would be based on
my wanted list. The fact that I have zamor spheres as one item on my list would
mean that all zamor spheres would be totaled and counted as one lot in the list.
Mark
|
|
|
| | | | | | | | | | | | | | | | | | | | | |
| | | | | | | | | | | | | Author: | mathematical | | Posted: | Aug 16, 2013 09:04 | | Subject: | Re: Improve Wanted List / Shop By | | Viewed: | 25 times | | Topic: | Suggestions | |
|
| In Suggestions, mhn1957 writes:
| | In Suggestions, mathematical writes:
| | Well...
If the program also factored in minimum buy (to reduce unneeded parts bought)
and shipping charges, the problem is actually a version of the traveling salesman
problem and is NP-hard, (Non-deterministic Polynomial-time hard) (You
can look both of those up on Wikipedia or something!). So, it is possible
for an implementation, but the time to solution cannot be solved in polynomial
time (read: reasonable time), unless P=NP (which has not been shown to be true,
but neither has P~=NP).
I don't mean to be mean or to burst your bubble or anything, but there's
just no way Bricklink would have enough computing power to accomplish this.
|
I think we could get closer to a solution by changing the way lots are matched,
or possibly adding a sequence/level of importance to the wanted lists or items
in them. To give an example, I added zamor spheres to my wanted list and that
one item destroyed any hope of getting a meaningful list of stores until I adjusted
the price down to $.03. On my wanted list, I set the price of things that are
secondary very low, or set the quantity very high, so that, after I find a store
that has most of the top tier items, I can press the (all items) button and get
a more comprehensive list. That's only after I've selected a couple of
stores to check. The wanted list maintenance is a little clunky, and I can think
of a number of ways using a 2 or 3 level search would work without attempting
a cpu intensive search which would bring up multiple stores as in the suggestion.
If I could specify that all zamor spheres should be considered one lot as far
as my search is concerned, that would help. The total number would be more important
to me than what colors, as I don't care about color in this instance. Since
they are listed by color, and some stores have multiple lots of a color, including
this item really distorts things.
In natural language,
Get a list of all stores that have the all of my level one items. (Required items.)
Take that list and display the stores, ranking them by the lot count and quantity
of level two items, based on my sort preference. Lot count would be based on
my wanted list. The fact that I have zamor spheres as one item on my list would
mean that all zamor spheres would be totaled and counted as one lot in the list.
Mark
|
Now that's the way to do it! 
I don't deny there are many better ways than perfect, but perfect here is
quite close to impossible.
|
|
|
| | | | | | | | | | | | | | | | | | | | | | | | | |
| | | | | | | | | | | | | | | Author: | mhn1957 | | Posted: | Aug 16, 2013 10:43 | | Subject: | Re: Improve Wanted List / Shop By | | Viewed: | 28 times | | Topic: | Suggestions | |
|
| In Suggestions, mathematical writes:
| | In Suggestions, mhn1957 writes:
| | In Suggestions, mathematical writes:
| | Well...
If the program also factored in minimum buy (to reduce unneeded parts bought)
and shipping charges, the problem is actually a version of the traveling salesman
problem and is NP-hard, (Non-deterministic Polynomial-time hard) (You
can look both of those up on Wikipedia or something!). So, it is possible
for an implementation, but the time to solution cannot be solved in polynomial
time (read: reasonable time), unless P=NP (which has not been shown to be true,
but neither has P~=NP).
I don't mean to be mean or to burst your bubble or anything, but there's
just no way Bricklink would have enough computing power to accomplish this.
|
I think we could get closer to a solution by changing the way lots are matched,
or possibly adding a sequence/level of importance to the wanted lists or items
in them. To give an example, I added zamor spheres to my wanted list and that
one item destroyed any hope of getting a meaningful list of stores until I adjusted
the price down to $.03. On my wanted list, I set the price of things that are
secondary very low, or set the quantity very high, so that, after I find a store
that has most of the top tier items, I can press the (all items) button and get
a more comprehensive list. That's only after I've selected a couple of
stores to check. The wanted list maintenance is a little clunky, and I can think
of a number of ways using a 2 or 3 level search would work without attempting
a cpu intensive search which would bring up multiple stores as in the suggestion.
If I could specify that all zamor spheres should be considered one lot as far
as my search is concerned, that would help. The total number would be more important
to me than what colors, as I don't care about color in this instance. Since
they are listed by color, and some stores have multiple lots of a color, including
this item really distorts things.
In natural language,
Get a list of all stores that have the all of my level one items. (Required items.)
Take that list and display the stores, ranking them by the lot count and quantity
of level two items, based on my sort preference. Lot count would be based on
my wanted list. The fact that I have zamor spheres as one item on my list would
mean that all zamor spheres would be totaled and counted as one lot in the list.
Mark
|
Now that's the way to do it!
I don't deny there are many better ways than perfect, but perfect here is
quite close to impossible.
|
Small improvements would be welcome. And thanks, Math person, I learned something
today. When I say learned, I mean became more aware of some things I don't
entirely understand.
Mark
|
|
|
| | | | | | | | | | | | | | | | | |
| | | | | | | | | | | Author: | RobErNat | | Posted: | Aug 16, 2013 11:01 | | Subject: | Re: Improve Wanted List / Shop By | | Viewed: | 22 times | | Topic: | Suggestions | |
|
| In Suggestions, mathematical writes:
| | Well...
If the program also factored in minimum buy (to reduce unneeded parts bought)
and shipping charges, the problem is actually a version of the traveling salesman
problem and is NP-hard, (Non-deterministic Polynomial-time hard) (You
can look both of those up on Wikipedia or something!). So, it is possible
for an implementation, but the time to solution cannot be solved in polynomial
time (read: reasonable time), unless P=NP (which has not been shown to be true,
but neither has P~=NP).
I don't mean to be mean or to burst your bubble or anything, but there's
just no way Bricklink would have enough computing power to accomplish this.
|
I am really dazzled...
Now open your store worldwide as mathematicly ( ) your are losing a lot of
potential sales
Nothing to do with P or NP, in worst case an NPB 
|
|
|
|
|
|
France, Normandie
Belgium, Flemish Brabant
USA, Nebraska
