hey everybody,
I’m trying to select the feature with highest linear correlation, in first pic u can see my data (mean, slope …), first colomn is:
1: Cooler condition / %:
3: close to total failure
20: reduced effifiency
100: full efficiency
by feature selecting I mean, I want to achieve the most correlated feature(mean, slope…)
I got the Linear correlation easily and my feature in this case (median) but I can not select the feature in order to use it in next step for my work, any idea? is it possible to use Linear C in feature selection node(loop)?!
1|690x350
You could adapt this workflow. You calculate a (Pearson) correlation between all numeric variables and the target variable (here: SalePrice). Then you use the absolute value to rank the variables and you could then cut at a number or value you want (negative values indicate a reversed correlation).
The housing price data is from a Kaggle competition.
kn_example_housing_prices_regression.knar (342.4 KB)
3 Likes
This topic was automatically closed 182 days after the last reply. New replies are no longer allowed.