Hi All,
One of my known source is predicting house prices given multiple independent variables.
In this case i would have used the old war house linear regression. However he is inclined towards using Random Forest. As I know these are advanced decision trees which reduces the issue of overfitting with other benefits.
But I am not sure whether we can use random Forest for such a prediction. Please correct me I am wrong. Any literature in case this claim is true would really help.
Dear Shivi Bhatia,
In my experience I have found Random Forest to be extremely useful for predicting for my regression problems. It is a highly flexible tool that is relatively tolerable to messy data. I found this paper was extremely useful for describing the use of the randomForest package in R: ftp://131.252.97.79/Transfer/Treg/WFRE_Articles/Liaw_02_Classification%20and%20regression%20by%20randomForest.pdf
There are plenty of examples in the literature of how people have used Random Forest models to predict: https://scholar.google.co.uk/scholar?hl=en&q=Random+Forets+predictions&btnG=&as_sdt=1%2C5&as_sdtp=.
Additionally, I have found that Boosted Regressions, which also base predictions on decision trees, perform similarly well to Random Forest analyses. Here is a nice paper describing their implementation in R:
ftp://131.252.97.79/Transfer/Treg/WFRE_Articles/Liaw_02_Classification%20and%20regression%20by%20randomForest.pdf
In summary I would recommend Random Forest as a highly flexible and efficient tool for prediction.
Yours Sincerely,
Joseph Crisp
Dear Shivi Bhatia,
In my experience I have found Random Forest to be extremely useful for predicting for my regression problems. It is a highly flexible tool that is relatively tolerable to messy data. I found this paper was extremely useful for describing the use of the randomForest package in R: ftp://131.252.97.79/Transfer/Treg/WFRE_Articles/Liaw_02_Classification%20and%20regression%20by%20randomForest.pdf
There are plenty of examples in the literature of how people have used Random Forest models to predict: https://scholar.google.co.uk/scholar?hl=en&q=Random+Forets+predictions&btnG=&as_sdt=1%2C5&as_sdtp=.
Additionally, I have found that Boosted Regressions, which also base predictions on decision trees, perform similarly well to Random Forest analyses. Here is a nice paper describing their implementation in R:
ftp://131.252.97.79/Transfer/Treg/WFRE_Articles/Liaw_02_Classification%20and%20regression%20by%20randomForest.pdf
In summary I would recommend Random Forest as a highly flexible and efficient tool for prediction.
Yours Sincerely,
Joseph Crisp
I too have gone through the article Joseph Crisp has shared above. It was first published in R news on Random Forests titled " Classification and Regression by
randomForest". It will help you as a great starter and let us know if you need further help. All the best...
Statistics is like an evolving process i feel. New methods new processes all the time. Thank you Ranga i will check and will let you know if i need further assistance.
Dear Shivi Bhatia,
I also think that Random Forests Regression can be used for your topic but you should keep in mind that a RF Regression is based on decision trees and thus can only deliver results within the borders of the training data. Extrapolation of your data as you could do with a linear regression model is as far as I know not possible.
Best regards, Philipp
- Badges
- Science topic
- Similar topics
- Medicine
- Physiology
- Biosignals
- Biomedical Signal Processing
- Statistical Analysis