Okay...Consider this problem.
A bank manager is carrying out a general customer survey concentrating primarily on customers' decision either to continue being associated with the bank or leave the bank altogether and seek alternative sources.
A questionaire has already been drafted and circulated.
Data collection is already done.
Now the data consists of say 200 data points (or 200 sets of observations corresponding to 200 different customers). Each observation accounts for data on say 50 different questions (predictor or independent or explanatory variables : 'X's) divided and clubbed under some 7 sections such as 'Service' , 'Incentives' etc. Also, data on each of these variables is a response on a 5-point scale (1-Strongly Agree, 2-Agree, 3-Neither Agree nor disagree, 4-Disagree, 5-Strongly Disagree). The results are already collected and partly analyzed as to which of these 200 customers intend to stay with the bank and which intend to leave the bank for some other source.
With the rest of the data, The main question now is to analyze :
1) Which are the top 5 of the 50 (positive) points that make a customer stay with the bank.
2) Which are the top 5 of the 50 (negative) points that make a customer leave the bank.
My initial guess was Discriminant Analysis. But that won't work coz it basically will tell the top factors that are responsible for discriminating between those who leave and those who stay. [We need different factors for those who stay (positive) and those who leave (negative).]
Secondly Factor Analysis is not an option since no data is to be lost in the process.
Can anyone suggest an approach to the problem? Its urgent so quick help will be appreciated.
A bank manager is carrying out a general customer survey concentrating primarily on customers' decision either to continue being associated with the bank or leave the bank altogether and seek alternative sources.
A questionaire has already been drafted and circulated.
Data collection is already done.
Now the data consists of say 200 data points (or 200 sets of observations corresponding to 200 different customers). Each observation accounts for data on say 50 different questions (predictor or independent or explanatory variables : 'X's) divided and clubbed under some 7 sections such as 'Service' , 'Incentives' etc. Also, data on each of these variables is a response on a 5-point scale (1-Strongly Agree, 2-Agree, 3-Neither Agree nor disagree, 4-Disagree, 5-Strongly Disagree). The results are already collected and partly analyzed as to which of these 200 customers intend to stay with the bank and which intend to leave the bank for some other source.
With the rest of the data, The main question now is to analyze :
1) Which are the top 5 of the 50 (positive) points that make a customer stay with the bank.
2) Which are the top 5 of the 50 (negative) points that make a customer leave the bank.
My initial guess was Discriminant Analysis. But that won't work coz it basically will tell the top factors that are responsible for discriminating between those who leave and those who stay. [We need different factors for those who stay (positive) and those who leave (negative).]
Secondly Factor Analysis is not an option since no data is to be lost in the process.
Can anyone suggest an approach to the problem? Its urgent so quick help will be appreciated.