I am stuck at 0.79166 in Loan Prediction Problem. I tried many things like making new variables. But still the result is same. What should I do?
@gau2112 - I would suggest you to try different models like random forest , boosting .Also try to perform ensembling which can improve the performance of your model.
Hope this helps!
I tried different models. I got the result with Random Forest. With others, i am getting low accuracy. Also How do I ensemble since i have to get a categorical variable in the end?
OK. I will post it as a new topic.
Hi gau2112, Which accuracy test have you used to test your model.
I did not use any accuracy test. In random forest, i only checked OOB estimate and confusion matrix .Can you suggest some accuracy tests or good articles on them?
Hi @gau2112 , I have used Logistic Regression modeling and considered the loan status as Y if the probablity is coming more than 0.5 or else N.Then i am checking how many Loan status i have predicted correctly in the “train data set” .What is your take on this?
I think it is fine. But I think logistic regression would be giving you less accuracy. What is your score on leaderboard?
Hi @gau2112 , I have done considerable efforts with future engineering but using logistic regression i am only getting an accuracy of 0.7847 .May be i should use another algorithm !!! Please do share which algorithm might give a better score than above if you are getting score more than 0.78477
I used random forest. You can also use xgboost. These two can give good accuracy.
I am new to this area, I am also trying for loan prediction model using logistic regression, some how I get only credit rating and property area as significant in the regression equation.
Tried to treat missing values of loan amount and loan term by using mean from this data. when i performed the check for the accuracy of the model, it gives me 0.
Can someone please guide me on the approach of fixing the problem.
change the loan_status to Y and N