These are chat archives for mohanmb91/Mohan_Febi_Avinash

21st
Apr 2016
Mohan Kumar Memangalam Balasubramani
@mohanmb91
Apr 21 2016 05:36
eric we have got the required sources and downloaded the data and parsed it now we have all different sources of data in one model
its a list of array of 2 crore records
now for cleaning the data we are planning to just loop through the models and finding the mode of the columns and replacing with null values and changing all the non numeric feilds into numeric feild
is this fine
??
Eric Liao
@rcliao
Apr 21 2016 05:47
ask yourself
Mohan Kumar Memangalam Balasubramani
@mohanmb91
Apr 21 2016 05:54
i think this is enough eric cause we are just facing few null values and i am planning to change it with mode or average of that particular column...
Febi Elgiva Kennedy
@febielgiva
Apr 21 2016 05:56
and also do we need to convert all non numeric fields to numeric fields in data mugee stage itself or we can do that while applying prediction algorithms.?
@rcliao
Eric Liao
@rcliao
Apr 21 2016 23:40
Mungee is your core logic, it is up to your domain and your problem
Therefore, I'd argue that is like a domain specific problem and I will not be answering those