The Methods For Determining the Type of Data Need Revisited.

[kvb2univpitt]

The method isMixed() of the BoxDataSet determines if the dataset is mixed by counting the number of continuous variables and discrete variables the dataset has. If the number of discrete variables and continuous variables are both non-zero, it is considered as mixed. This is not quite right, because the max-discrete-category value can be set small such that there will be no variable considered discrete. In this case, the dataset is still mixed, but the method isMixed() will return false and the method isContinuous() will return true. Having said that, the method isContinuous() and isDiscrete() are not correct either.

[kvb2univpitt]

@jdramsey Joe, what are your thoughts?

[yuanzhou]

Just want to mention that in algo chooser we are also using isContinuous(), isDiscrete(), or isMixed() to decide the applicable algorithms based on the data type of datasets.

[jdramsey]

@kvb2univpitt I think if you set the max discrete value too low and it reads it in as continuous, it's your fault. :) The dataset you read in is in fact continuous.

Or are you suggesting that if a variable has only two values, it should be interpreted as discrete even if the values are all real values? That's a possibility.

Sorry guys for not checking all of these issues earlier!

I think this issue can be closed. :)

[cg09]

Two valued variables with real values should NOT be treated as discrete. If they were, then binary variables could never be used in a linear regression.

…

On Tue, Apr 9, 2019 at 1:19 PM Joseph Ramsey ***@***.***> wrote: @kvb2univpitt <https://github.com/kvb2univpitt> I think if you set the max discrete value too low and it reads it in as continuous, it's your fault. :) The dataset you read in is in fact continuous. Or are you suggesting that if a variable has only two values, it should be interpreted as discrete even if the values are all real values? That's a possibility. Sorry guys for not checking all of these issues earlier! *I* think this issue can be closed. :) — You are receiving this because you are subscribed to this thread. Reply to this email directly, view it on GitHub <#888 (comment)>, or mute the thread <https://github.com/notifications/unsubscribe-auth/APmNuYpCO823LAB8yJL8oe0FyQY6E9yFks5vfMu2gaJpZM4V-kI_> .

[jdramsey]

Good point.

…

On Tue, Apr 9, 2019 at 2:03 PM cg09 ***@***.***> wrote: Two valued variables with real values should NOT be treated as discrete. If they were, then binary variables could never be used in a linear regression. On Tue, Apr 9, 2019 at 1:19 PM Joseph Ramsey ***@***.***> wrote: > @kvb2univpitt <https://github.com/kvb2univpitt> I think if you set the > max discrete value too low and it reads it in as continuous, it's your > fault. :) The dataset you read in is in fact continuous. > > Or are you suggesting that if a variable has only two values, it should be > interpreted as discrete even if the values are all real values? That's a > possibility. > > Sorry guys for not checking all of these issues earlier! > > *I* think this issue can be closed. :) > > — > You are receiving this because you are subscribed to this thread. > Reply to this email directly, view it on GitHub > <#888 (comment)>, > or mute the thread > < https://github.com/notifications/unsubscribe-auth/APmNuYpCO823LAB8yJL8oe0FyQY6E9yFks5vfMu2gaJpZM4V-kI_ > > . > — You are receiving this because you were mentioned. Reply to this email directly, view it on GitHub <#888 (comment)>, or mute the thread <https://github.com/notifications/unsubscribe-auth/AJZZR1zw_7RNqtza8UV4tXP7PXAmSSK-ks5vfNXwgaJpZM4V-kI_> .

-- Joseph D. Ramsey Special Faculty and Director of Research Computing Department of Philosophy 135 Baker Hall Carnegie Mellon University Pittsburgh, PA 15213 jsph.ramsey@gmail.com Office: (412) 268-8063 http://www.andrew.cmu.edu/user/jdramsey

[jdramsey]

This all works now and has for a long time. Closing.