Hi,
less comlex (?...) but less complete : use fuzzdemo from LFTP.
Using a similarity distance function (y=x), you'll be able to define
shortest distance to a store, mean distance to all stores for a customer,
and /or the equivalent for each store ... But they won't be linked ...
You can use a more complex function, to better qualify the distance between
a store and a customer (for example introducing a cost by the mean of a
differential function - ie : altitude differential, or anything you can
think of ...)
BTW: I hadn't have much news from this application since I posted it. Very
very few people reported using it :( .
These days, a few (very few, but ...) are popping up ... So please give me
feed back, I'm not going to investigate further in this direction (fuzzy
logic ...) if it doesn't appear like something useful.
Right now I have to make a decision as to an evolution of it or just keep
on using it only for my own use (scarce since MI and GIS are not my main
activity stream ...)
So, if anyone'd like me to support this tool, let me know ... or please be
kind enough to give me feedback ...
Have a nice day.
Eric.
> -----Message d'origine-----
> De: Bill Thoen [SMTP:[EMAIL PROTECTED]]
> Date: mercredi 20 octobre 1999 19:55
> A: [EMAIL PROTECTED]
> Cc: MapInfo-L
> Objet: MI How to join customers to nearest stores
>
> Associating customers with the nearest stores is do-able, but a
> bit complicated. This ought to be a MapInfo built-in function,
> because if you are new to MapInfo, most of the following probably
> will go right past you, but it does work. I've cc'ed this to the
> MapInfo-L list so maybe someone who has a bit of time could turn
> this into a generic utility for new MapInfo users. And perhaps
> someone can come up with a simpler approach.
>
> Anyway, here goes:
>
> The general object here is to create a table that has all your
> customers and their distances to all the stores. You order that
> file by customer ID and Distance. Then you take that table and
> group by customer ID (since the shortest distance to any of the
> stores was listed first, the group by clause will grab this first
> record for each customer-store record. All you need from this
> last file is store Id and customer Id, and then save that table
> (so it will be a permanent table, because we need to use it in a
> join next.) Open the table you just saved and join it to the
> customer records. Now for each customer, you have the store id
> for the store that is the closest to that customer.
>
> To create the starting table with all stores and customers with
> their distances, you first need to add some columns to each
> table. To the Stores table, add three fields with the following
> names and types: StoreLon float, StoreLat float, and Link
> integer. To the Customers table, add three fields with the
> following names and types: CustLon float, CustLat float, and Link
> integer. Update the tables to fill in the points' coordinates
> into the lat and lon fields (using the CentroidX and CentroidY
> functions. In the end you should have the coordinates for the
> stores in StoreLon, StoreLat, the coordinates for Customers in
> CustLon and CustLat, and leave 0 in both Link fields.
>
> No do a cartesian join on your two tables using the Link field.
> Join the tables where the link field in one matches the link in
> the other. Of course these are all zero, so you'll get a table
> with every customer joined to every store (which is what we
> want.) In the list of fields that you select, you will also need
> to create a calculated field for distance, so the full SQL select
> might look like:
>
> Select stores.StoreId, Customers.CustId,
> Distance(stores.StoreLon, stores.StoreLat,
> Customers.CustLon, Customers.CustLat, "mi") "Distance"
> from stores, Customers
> where stores.Link=Customers.Link
> order by Customers.CustId, Distance
> into JOIN
>
> Next SQL select that table (JOIN) and group by the CustId and
> create the table NEAREST, and extract only the Id numbers, like
> so:
>
> Select StoreId, CustId
> from JOIN
> group by CustId
> order by CustId
> into NEAREST
>
> Now save this table; close the temporary version and open the one
> you just saved so that now you have a permanent table named
> Nearest. Join this with the Customer table like so:
>
> Select Customers.CustId, Customers.Name, Nearest.StoreId
> from Customers, Nearest
> where Customers.CustId=nearest.CustId
> order by Customers.CustId
> into NEAREST2
>
> Now you have a list of customers and their names (in this case)
> with the store Id nearest to them.
>
> Simple, huh?
>
> - Bill Thoen
>
>
> Berk Charlton wrote:
> >
> > Mapinfo tech support has been unable to help me with this scenario:
> >
> > Let's say I have two point tables, one is a table of store locations,
and the
> > other is a table of customers. I need to be able to put the store id
and
> > distance to the closest store to each of the customer points.
> >
> > I new to Mapinfo, but in the past been able to do this in Arcview and
Atlas GIS
> > easily.
> >
> > Any suggestions?
> >
> > Thanks,
> > Berk Charlton
> ----------------------------------------------------------------------
> To unsubscribe from this list, send e-mail to [EMAIL PROTECTED] and put
> "unsubscribe MAPINFO-L" in the message body, or contact [EMAIL PROTECTED]
----------------------------------------------------------------------
To unsubscribe from this list, send e-mail to [EMAIL PROTECTED] and put
"unsubscribe MAPINFO-L" in the message body, or contact [EMAIL PROTECTED]
