Here is a dataset:

Position Level Performance Previous Promotion Promotion

Entry High No Yes

Mid$-$level High Yes Yes

Entry High No Yes

Entry Low No Yes

Entry Low No No

Mid$-$level Low Yes No

Mid$-$level High No Yes

Mid$-$level Low Yes No

Entry Low Yes Yes

I calc the overall Gini index for this:

Promotion

Proportion $($No$)=3/9=0\mathrm{.}33$

Proportion $($Yes$)=6/9=0\mathrm{.}67$

Gini Index $=0\mathrm{.}33(10\mathrm{.}33)+0\mathrm{.}67(10\mathrm{.}67)$

Gini Index $=0\mathrm{.}33(0\mathrm{.}67)+0\mathrm{.}67(0\mathrm{.}33)$

Gini Index $=0\mathrm{.}2211+0\mathrm{.}2211$

Gini Index $=0\mathrm{.}4422$

To make the next split, I calc all Gini indices for $2$ subregions R$1/$R$2$:

Weighted average for Position Level $=0\mathrm{.}3404$

Weighted average for Performance $=0\mathrm{.}3244$

Weighted average for Previous Promotion $=0\mathrm{.}3404$

So$,$ then I know the split should be on Performance. The next split, I calc all Gini indices:

Weighted Previous Promotion $=0\mathrm{.}3564$

Weighted Position $=0\mathrm{.}3670$

So$,$ the next split is Previous Promotion. Which leaves one more split, and Gini calc is:

Weighted Position Level $=0\mathrm{.}3298.$

The Gini index overall $0\mathrm{.}4422>0\mathrm{.}3298$ after $4$ regions$/3$ splits. Meaning I made an improvement. Is that correct?

Also, how can I draw the decision tree in a tree$-$based format? That is$,$ draw the decision tree and each split, and indicate which prediction we would make for each region? Please help me$.$ Thanks.