Preference Value

["F.Ozgur Catak" <f....@gmail.com>]

Hi,

When we prepare our data, we need 3 information. User Id, Item ID and Pref
Value. Why we use Pref Value. What does it mean? When we use similarity,
does it need to be used?

Sincerely

*Ozgur CATAK
**Istanbul University, PhD Student.*

Re: Preference Value

[Sean Owen <sr...@gmail.com>]

Yea you can do that. The framework specializes for this case though. See the
'boolean' data structures and classes. For example when using file input,
just omit the pref value on each line and it will just work.

Note that several components don't make sense without press, like
Pearson-based metrics. Stuff like log likelihood works, however, and is a
lot faster.

On Dec 27, 2009 9:00 PM, "Ted Dunning" <te...@gmail.com> wrote:

If your users rate items, then preference value should be the rating that
they give the item.  If you are using an implicit observation such as
clicking or viewing, then you should use 1 for the preference value.

On Sun, Dec 27, 2009 at 9:44 AM, F.Ozgur Catak <f.ozgur.catak@gmail.com
>wrote:

> > When we prepare our data, we need 3 information. User Id, Item ID and
Pref > Value. Why we use ...

Re: Preference Value

[Ted Dunning <te...@gmail.com>]

If your users rate items, then preference value should be the rating that
they give the item.  If you are using an implicit observation such as
clicking or viewing, then you should use 1 for the preference value.

On Sun, Dec 27, 2009 at 9:44 AM, F.Ozgur Catak <f....@gmail.com>wrote:

>
> When we prepare our data, we need 3 information. User Id, Item ID and Pref
> Value. Why we use Pref Value. What does it mean? When we use similarity,
> does it need to be used?
>
>