Chaid model result always return 0 accuracy about chefboost HOT 7 CLOSED

Would you please share dataset?

from chefboost.

I can not share all data because of corporate company data security problems but I can give details about form of data.

Independent columns are like below respectively.
INVTRF_L1L2L3_1_2_FLG_OBIN (int64)
MAX_GECIKME_L6M_OBIN (int64)
MAX_SUBSCRIBER_TENURE_MONTHS_OBIN (int64)
SUBS_POSTPAID_TENURE_DAYS_OBIN (int64)
TRF_PRICE_CHNG_RATIO_L12M_OBIN (int64)

Target variable : Decision (int64)

Top 5 data is like this:

0 3 7 0 1 6 0
1 1 8 4 4 4 1
2 3 6 9 2 6 0
3 3 0 12 4 5 0
4 2 6 6 4 4 1

Ordinal transformation was applied to columns and all data is in numeric form.

from chefboost.

oh if the column types are actually classes instead of real numeric values, then you should make them string.

e.g. instead of 0, make it class_0.

Otherwise, chaid will build trees with rules being greater or less than a threshold value as explained in this video: https://www.youtube.com/watch?v=YTC7sTXDmWk

from chefboost.

The problem was solved, the model worked thank you so much :)

How can I visualize the tree ?

from chefboost.

trees cannot be visualized with chefboost, you can just check the output .py files out.

from chefboost.

This is so sad. Do you know any other lib for visualization using rules.py file?

from chefboost.

nope

from chefboost.