ross classification rules

DA CW 2

Professions had typos: below are the changes made to correct data.
Ather > other
busines > business
busness > business
Radio/Tv > radio/tv
Eduction > education

No duplicates found.
All incorrect/unrealistic data was removed. Screenshots were taken.

Used J48 to find classification rules as it showed the dataset in a parent/child tree format. This was easier to interpret than the likes of OneR where the data is presented in a irrelvant fashion. J48 is an improbed version of ID3 which allows the use of numbers aswell as text. Due to my dataset containing both numbers and text this was neccesary.

Classification Rules:

IF checking_status = <0
AND credit_history = existing paid
AND saving_status = no known savings
AND job = unskilled resident
THEN class = good
*3 instances

IF checking_status = <0
AND credit_history = existing paid
AND saving_status = <100
AND purpose = radio/tv
AND employment = 1<=X<4
AND job = high qualif/self emp/mgmt
THEN class = good
*2 instances

IF checking_status = 0<=X<200
AND saving_status <100
AND credit_amount <= 9283
AND purpose = new car
AND employment = >=7
THEN class = bad
*5 instances

IF checking_status = 0<=X<200
AND saving_status <100
AND credit_amount <= 9283
AND purpose = business
AND employment 1<=X<4
AND age >29
THEN class = bad
*3 instances

IF checking_status = 0<=X<200
AND saving_status <100
AND credit_amount <= 9283
AND purpose = business
AND employment = 4<=X<7
THEN class = good
*2 instances

IF checking_status = 0<=X<200
AND saving_status <100
AND credit_amount > 9283
THEN class = bad
*12 instances (Jordan has 13)

IF checking status = <0
credit_history = existing paid
saving_status >=1000
THEN class = good
*4 instances

IF checking_status = <0
Credit_history = existing paid
Saving status = <100
Purpose = furniture/equipment
Employment >=7
Personal_status = male single
THEN class = bad
*4 instances (jordans)

IF checking_status = <0
Credit_history = existing paid
Saving status = <100
Purpose = furniture/equipment
Employment >=7
Credit amount >1680
Class = bad (2 instances)
if credit amount <=1680
class = good (2 instances)