dozercodes / decisiontree Goto Github PK

https://github.com/NinjaSteph/DecisionTree/blob/18bb6a6bcc216f3682b40540aaa678b6b2e41ee9/sna2111_DecisionTree/DecisionTree.py#L42

i+=1

In entropy function, you should use "i+=1" instead of "++1".
In choseAttr add "if attr != target" after the "for", So the algorithm wont choose the target class.

At No.42 row in DecisionTree.py ++i can NOT be used for increasing i step by step

There is an error in the entropy calculation so the tree generated is wrong by the ID3 code.
I modify the loop at row 38 of DecisionTree.py in this way
i = 0
for entry in attributes:
if (targetAttr != entry):
i+=1
else:
break
if you modify that there is another bug on the ChooseAttrib function, it will select the highest gain value attrib, that is the target, just remove the target from the list of the attributes.
Nice code

At No. 90 row in DecisionTree.py, when choosing the best attribute, there exists trouble traversing the whole attributes. The following code is advised to add in the for loop:
if attr == target:continue