gave another tip for storing the matrix using an adjacent table,
coupled with many to many.
that's just how it is implemented in classes django,
class Answer (models.Model):
id_a = models.AutoField ('ID', primary_key = true)
answer = models.TextField ()
class Simptoms (models.Model):
id_s = models.AutoField ('ID', primary_key = true)
simptom = models.TextField ()
class Answer_Simptoms (models.Model):
id_a = models.ManyToManyField (Answer)
id_s = models.ManyToManyField (Simptoms)
But I think where I nakosyachil.
although, in principle, what I should be.
On 9 апр, 21:32, Bill Freeman <ke1g...@gmail.com> wrote:
> Will there be more than one matrix? Or are you using "Name" to indicate
> matrix membership, as opposed to column names and row names? If
> so, a foreign key on a matrix table will be more storage efficient, since
> matrix names will only need to be stored once.
>
> Also, doesn't a select to do a lookup of a data item always need to join
> on all three tables? If you want to go this route, then you only really need
> one table, though, if there are multiple named matrices, a second table
> to hold the matrix name would be more efficient.
>
> Let's say, for the sake of argument, that are numbering your matrices,
> rows, and columns, rather than naming them, and avoid the extra table(s)
> and join(s).
>
> create table Data
> (
> id int primary key,
> matrix int not null,
> row int not null,
> col int not null,
> data float not null
> )
>
> You can query this for an exact data item by specifying all of matrix, row,
> and
> col in your where clause. Or you can just specify matrix and row and get
> all of a col (perhaps order by col) to get one row. Similar for
> getting a column.
>
> In my initial scheme, the Data table items, which greatly outnumber any of
> the other table's items, were slightly smaller each, having an id, a col int,
> and a data (of whatever type). This has the cost that Data must be joined
> on Row, in turn joined on Matrix if matrices have a name, and getting one
> column is a more expensive operation (but still done in one where clause).
>
> The speed versus space tradoff is yours to make.
>
> Also, while it may be easier for you to think in terms of SQL, to wire this to
> Django, you should probably write it as a model rather than as create
> table SQL, so that you will see when you are going somewhere difficult.
>
> For my first scheme above (remember, id is automatic):
>
> class Data(models.Model):
> matrix = models.IntegerField()
> row = models.IntegerField()
> col = models.IntegerField()
> data = models.FloatField()
>
> Now you can, for instance, do something like:
>
> for d in Data.objects.filter(matrix=m, row=r).order_by('col'):
>
> to iterate over a row in order (leave off the order_by if order doesn't
> matter),
> and:
>
> d = Data.objects.get(matrix=m, row-r, col=c)
>
> to get an individual data object, which you can update using:
>
> d.data = new_value
> d.save()
>
> use filter instead of get if you want to allow a matrix to be sparse.
>
> I'll leave the three (or two, if matrices are numbered) table scheme as
> an exercise for you.
>
> Bill
>
>
>
> On Fri, Apr 9, 2010 at 3:50 AM, zimberlman <zimberl...@rambler.ru> wrote:
> > create table Rows
> > (
> > Id int primary key,
> > Name text(255) not null
> > )
> > create table Cols
> > (
> > Id int primary key,
> > Name text(255) not null
> > )
>
> > create table Data
> > (
> > RowId int not null foreign key references Rows(Id),
> > ColId int not null foreign key references Cols(Id),
> > Data
> > )
>
> > On 8 апр, 02:32, Bill Freeman <ke1g...@gmail.com> wrote:
> >> You need one matrix table, having a row for each matrix.
>
> >> You need one matrix_row table, having a row for each row of any matrix,
> >> mostly
> >> containing a foreign key on the matrix table, showing of which matrix the
> >> row is
> >> part, plus it's row number in that table.
>
> >> And you need one matrix_row_values table, having a foreign key on the
> >> matrix_row,
> >> indicating of which row it is a part, plus it's column number, and the
> >> value.
>
> >> There is no run time table creation involved, and this can all be done
> >> with direct
> >> django models, and on any database the ORM supports (maybe not NoSQL
> >> databases, I'll have to think about that). The access syntax wouldn't
> >> look like
> >> 2D array access, but you could fix that with a wrapper.
>
> >> On Wed, Apr 7, 2010 at 4:20 PM, zimberlman <zimberl...@rambler.ru> wrote:
> >> > No, no.
> >> > I need to store the matrix and dual array is ideal for this would come
> >> > up.
> >> > The problem is that the matrix will grow in size, thousands of entries
> >> > only on i, and where that same number of j.
> >> > create table is not an option.
>
> >> > only if the matrix transform and drive the table. and my question is,
> >> > if you can not define arrays in the model, how can I convert a
> >> > matrix.
> >> > example is the matrix
> >> > a b c d e
> >> > 1 2 3 4 5
> >> > 11 22 33 44 55
> >> > converted into
> >> > a 1
> >> > a 11
> >> > b 2
> >> > b 22
> >> > c 3
> >> > c 33
> >> > d 4
> >> > d 44
> >> > e 5
> >> > e 55
>
> >> > this is for example, how do I implement this in the model? course in
> >> > the class will have to use functions, but is it possible?
>
> >> > On 8 апр, 01:28, pmains <peter.ma...@gmail.com> wrote:
> >> >> If there is no Django model field for this, then one option would be
> >> >> to create your own model field (http://docs.djangoproject.com/en/dev/
> >> >> howto/custom-model-fields/). Of course, it would not be compatible
> >> >> with most SQL Database types.
>
> >> >> Of course, it may be easier to just rethink your data model. Would it
> >> >> be possible to create one or more additional tables to avoid using an
> >> >> unusual DB construct like a 2D array?
>
> >> >> Could you explain the problem in a little more detail? Saying that you
> >> >> want a 2-dimensional array is very abstract, but if we understand what
> >> >> problem you are trying to solve, what data you are attempting to
> >> >> model, it will be easier to give you useful suggestions.
>
> >> > double array that I want to keep count, it is ideally suited for the
> >> > incidence matrix of the graph.
> >> > required to store the connection to the knowledge base of question-
> >> > answering system.
> >> > which will lie on j IDs leading questions, and i will lie on
> >> > identifiers answers emerging from a set of leading questions.
> >> > and a double array is an ideal way to store data.
> >> > but I probably need to write your own model of the array for django in
> >> > this case, it is the only big problem at this time.
>
> >> > --
> >> > You received this message because you are subscribed to the Google
> >> > Groups "Django users" group.
> >> > To post to this group, send email to django-us...@googlegroups.com.
> >> > To unsubscribe from this group, send email to
> >> > django-users+unsubscr...@googlegroups.com.
> >> > For more options, visit this group
> >> > athttp://groups.google.com/group/django-users?hl=en.
>
> > --
> > You received this message because you are subscribed to the Google Groups
> > "Django users" group.
> > To post to this group, send email to django-us...@googlegroups.com.
> > To unsubscribe from this group, send email to
> > django-users+unsubscr...@googlegroups.com.
> > For more options, visit this group
> > athttp://groups.google.com/group/django-users?hl=en.
--
You received this message because you are subscribed to the Google Groups
"Django users" group.
To post to this group, send email to django-us...@googlegroups.com.
To unsubscribe from this group, send email to
django-users+unsubscr...@googlegroups.com.
For more options, visit this group at
http://groups.google.com/group/django-users?hl=en.
