logistic regression operator and weights

Telcontar120

The operator information for the base logistic regression learner indicates it does not accept weighted examples (see screenshot). However, if you actually test this by running a model on weighted vs unweighted examples, it is very clear that the resulting model is different, so it does appear that weighting is affecting this operator. See the example process here:

<描述一致= =“绿色”“左”颜色色= " true" height="80" resized="true" width="248" x="37" y="137">In the training phase, a model is built on the current training data set. (90 % of data by default, 10 times)

















The model created in the Training step is applied to the current test set (10 %).<br/>The performance is evaluated and sent to the operator results.

A cross-validation evaluating a decision tree model.













<描述一致= =“绿色”“左”颜色色= " true" height="80" resized="false" width="248" x="37" y="137">In the training phase, a model is built on the current training data set. (90 % of data by default, 10 times)

















The model created in the Training step is applied to the current test set (10 %).<br/>The performance is evaluated and sent to the operator results.

A cross-validation evaluating a decision tree model.



<连接from_op = "过滤器示例”from_port = "穰mple set output" to_op="Multiply" to_port="input"/>

<连接from_op = "用" from_port = "输出2”to_op="Generate Weight (Stratification)" to_port="example set input"/>

Can the operator information be updated or clarified? Thanks.

sgenzer

Got it@Telcontar120. Pushing to dev team. Stay tuned.

sgenzer

Confirmed as issue. Ticket created.

Telcontar120

Hi@sgenzerI am not entirely sure why this one is classified as a feature request. It seems based on the initial analysis that the current operator is not working as specified (e.g., it says it does not accept weights but clearly it is doing something different when weights are present). So wouldn't this one be more of a bug fix? It could in theory be resolved in two very different ways:

• actually make it ignore the weights (as it says it does, but currently doesn't)
• or, actually use the weights, and update the operator capabilities description accordingly (note: this is clearly the better option!)

But the way this is framed currently, I am not sure what we would be getting if we voted for this issue. Can you clarify?