I've been assessing this alteryx capability today and its more cumbersome once you lift the lid a little. For example... the tool makes it easy to replace with median for missing values which makes a lot of sense however the way you assigned the median in the test data set is really replacing the nulls with the median of the test set. If you wanted to use the median from the training dataset, which I think is a more logical way to do this, the number of tools involved grows considerably because you can't just use the imputation tool and this should be something the intelligence tools should be able to incorporate seeing as they have already calculated this. Also I haven't proven it out but I think Passenger ID is unlikely to be a good predictor even though the tool can find some correlation.
I've been assessing this alteryx capability today and its more cumbersome once you lift the lid a little. For example... the tool makes it easy to replace with median for missing values which makes a lot of sense however the way you assigned the median in the test data set is really replacing the nulls with the median of the test set. If you wanted to use the median from the training dataset, which I think is a more logical way to do this, the number of tools involved grows considerably because you can't just use the imputation tool and this should be something the intelligence tools should be able to incorporate seeing as they have already calculated this. Also I haven't proven it out but I think Passenger ID is unlikely to be a good predictor even though the tool can find some correlation.
Why did you write Linear Regression in description? Isn't it logistic regression?
Good work, wish you could upload more
how much does ML in Alteryx cost?