[Tundra.L]

Up

0

Down

If this is what you mean:

normally I used following methods to assign labels for segments

1. customer function to define the segment, the labels will be numeric data

2.cut() or qcut (), the labels will be category data

3. ceil() for special situation, the labels will be numeric data

[Tundra.L]

Up

0

Down

My finding is: for both best and worst tier, 0 group’s debtCon/homelep ratio is always higher than 1 group, and the trends is almost the same cross those 10 tiers(decile)

[Tundra.L]

Up

0

Down

thank you

[Tundra.L]

Up

0

Down

thank you.

non-equal variance means we should use:

equal_var=False

correct?

[Tundra.L]

Up

0

Down

can you show us please?

[Tundra.L]

Up

0

Down

thank you

[Tundra.L]

Up

0

Down

thank you. I switched to loc() function, now it works. the output is like:

0 8535
1 1954
0 8496
1 1749
.......

• This reply was modified 4 years ago by  Tundra.L.

[Tundra.L]

thank you

[Tundra.L]

Up

0

Down

I misunderstand then. the TN is supposed be 1 & 1. I thought it is 1 &(1,0), that’s why

thank you

[Tundra.L]

Up

0

Down

got it. thank you

[Tundra.L]

Up

0

Down

thank you.

this is the qcut function I used:

cr_label=range(1,11)

Crdt_seg=pd.qcut(Loan[‘CR_Score’],q=10,labels=cr_label)

Loan=Loan.assign(credit_seg=Crdt_seg.values)

Loan.groupby(‘credit_seg’).agg({‘Client_ID’:’count’})

the output is little different about client numbers in each segments compare with yours. Is there anything wrong in my code?

• This reply was modified 4 years ago by  Tundra.L.
• This reply was modified 4 years ago by  Tundra.L.