How to make a MLR model from given datasets

jruth

New Member
#1
I'm sure this is easy to do, but I am having a hard time finding good online resources.
((This is not for a graded class assignment or exam, just assigned by my boss to become familiar with MLRs))
I have a variable Y (length 99) and 10 variables X1-X10 to build a regression model.
Basic correlations with each variable and the dependent output Y show no correlations stronger than .15 and within 'independent' variables no stronger than .22.
While my boss claims that an R^2 value greater than .85 is possible and has been done before with this dataset, I am not sure how to achieve a linear regression equation that approaches anything close to this. - attempts have only achieved R^2=.54 and adjusted R^2 of .49

If anyone can help me or recommend a site that is not https://www.statmethods.net/stats/regression.html or http://r-statistics.co/Linear-Regression.html that would give a good procedure for doing something like this I would appreciate it!!
 

Karabiner

TS Contributor
#2
Maybe your boss used expressions which accounted for non-linear relationships (e.g. he added x^2 or x^3 terms), and interactions of predictors?


With kind regards

Karabiner
 

jruth

New Member
#3
Maybe your boss used expressions which accounted for non-linear relationships (e.g. he added x^2 or x^3 terms), and interactions of predictors?


With kind regards

Karabiner

Thanks for the advice. I should have said- he did specify that we would only work with addition/subtraction, ratios of variables, or natural log of variables