I did not understand credit_history variable. Can any one help me with the explanation please?

loan_prediction

#1

Explanation needed for credit_history varible please.


#2

@pridhvi,
A credit history is a record of a borrower’s responsible repayment of debts. A credit report is a record of the borrower’s credit history from a number of sources, including banks, credit card companies, collection agencies, and governments.
In short it is the credibility of the applicant extracted from various sources.
Regards,
Syed Danish


#3

sir, thank you it was helpful but how to handel the missing values in this varible


#4

There are various methods to do that :

1.Drop these entire rows
2.Impute with any constant
3.Impute with mode
4.Run a model to predict the missing values
5.Create a new variable, as flag to indicate missing or non-missing


#5

sir, i have created the model on the training data set, now i got the probabilities, if i want to predict the loan status in the test data set i need to give certain limt to the probabilities right , what limit should i give. i mean should i say people who got more than 50% are elgible for loan and rest are not elegible is this the way or i am thinking in the wrong way


#6

Hi @pridhvi,

For setting up a threshold value , you must keep the business side of the problem in mind. For a bank, it is highly important not to give a loan to a person who does not deserve it. It may not be a big problem to them if they deny a loan to a deserving candidate. Thus from a bank’s perspective, false positives need to be avoided preferably.

Hence, they must keep the threshold above 0.5 for sure. This will ensure that a person even with a probability of 0.5 does not receive the loan and this makes it safer for the bank.

Now to determine the exact number of the threshold value, you can either make a calculated guess, or use the ROC curves to do the same.

Regards,
Shashwat


#7

Can someone explain me if more the Credit History better is one’s Credit Score? or Viceversa?


#8

I understand the credit_history variable to be whether a person’s credit history meets ‘guidelines’. I imagine for Dream Housing Finance, it has its own criteria for what a ‘good’ credit history is, and the 0 and 1 is binary. Meaning, 0 = false (not good credit history) and 1 = true (good credit history).

Hope that helps.


#9

Hi Danish,

credit history is an integer. How can we impute mode for this variable? Mode can be used for categorical variables. Please suggest whether we should use Mean/Median for this credit history variable?

Thanks,
Ritika


#10

According to me, a person can have a credit history either good (1 in this case) or bad (0 in this case). However, it is also possible that the person don’t have a history, because she never took any loan previously. Therefore, I am treating the variable as categorical with values 1, 0 and 2 (which are null, thus implying that they don’t have history.)


#11

I want to treat NA values of Credit history variable as new factor level.Can somebody help me with code?


#12

@BhaskarBiswas, I don’t think that is the case here. People who don’t have credit history can’t get a loan. So if the got a loan, I think we should assume they have a credit history. We should try to predict whther it was good or bad.


#13

Agreed, I also feel that there should be three category and there are many who have taken the first loan (that is no credit history)