How to determine whether to remove a generated outlier or not in stepwise regression?
1 Ansicht (letzte 30 Tage)
Ältere Kommentare anzeigen
Ajay Goyal
am 6 Jan. 2016
Kommentiert: Greg Heath
am 13 Jan. 2016
Dear Friends, I have developed a stepwise regression model. I am getting a outlier. I need to know the effective weight of the outlier so that I came to understand weather I need to remove it or my model will play good without removing it. In summary, I want to calculate adjusted R2 value with and without outlier.Can you please help
0 Kommentare
Akzeptierte Antwort
Greg Heath
am 6 Jan. 2016
My approach:
1. Use zscore (you can also use mapst) to standarize all variables to
zero-mean/unit-variance.
2. Use minmax to find the extrema of each variable
3. Consider all variables outside of [ -3, 3 ] as being outliers which
may have to be removed or modified.
4. Plot all variables which have outliers
5. Look at the plots and use judgement regarding one of the following choices:
a. leave it in
b. reduce the absolute value to 3 or another value of choice
c. remove it
6. When reporting the results
a. Explain the process
b. Compare the result with one or more of the considered alternatives.
Hope this helps.
Thank you for formally accepting my answer
Greg
2 Kommentare
Greg Heath
am 13 Jan. 2016
I prefer that
1. You contaminate an example
help nndatasets
2. Attempt to code and apply my algorithm.
3. Post your results
I will comment on your result.
Greg
Weitere Antworten (0)
Siehe auch
Kategorien
Mehr zu Descriptive Statistics finden Sie in Help Center und File Exchange
Community Treasure Hunt
Find the treasures in MATLAB Central and discover how the community can help you!
Start Hunting!