Closed dvquy13 closed 4 years ago
H: Hypothesis C: Conclusion I: Idea A: Assumption R: Remark
[x] H1. age takes 5 unique values in range [1,5]. They are probably the bins which abstract the real age such as 18, 25, 34, ...
age
[x] C2. Can ignore zipcodeOr and zipMerchant because they carries no information with only 1 unique value each.
zipcodeOr
zipMerchant
[ ] I3. Behavior at merchant are the richest and most potential source of information -> should focus to extract information.
merchant
R4. category is very skewed towards "es_transportation" (85%). The differentiation power is hard to tell but this feature can complement merchant
category
R5. amount is the only true continuous variable here and there are clear outliers which are very relevant in the context of fraud.
amount
[x] I6. Is there any merchants with disproportionate female customers?
[ ] I7. Use LRFMP features to describe customers
First assessment looking at the features one by one
H: Hypothesis C: Conclusion I: Idea A: Assumption R: Remark
[x] H1.
agetakes 5 unique values in range [1,5]. They are probably the bins which abstract the real age such as 18, 25, 34, ...[x] C2. Can ignore
zipcodeOrandzipMerchantbecause they carries no information with only 1 unique value each.[ ] I3. Behavior at
merchantare the richest and most potential source of information -> should focus to extract information.R4.
categoryis very skewed towards "es_transportation" (85%). The differentiation power is hard to tell but this feature can complementmerchantR5.
amountis the only true continuous variable here and there are clear outliers which are very relevant in the context of fraud.[x] I6. Is there any merchants with disproportionate female customers?
[ ] I7. Use LRFMP features to describe customers