Employee performance assessment using modified decision tree

Abstract EN

Decision tree algorithms are famous method in inductive learning and successfully applied for model classification and prediction.

Performance evaluation in organization is one of the most important issues that are reliable due to the transition from industrial to knowledge age.

This paper proposes the use of modified ID3 (Interactive Dichotomiser 3) decision tree algorithm combining with Taneja entropy instead of the original ID3 algorithm that depends on Shannon entropy which is widely used in the information theory.

In fact, the original ID3 was suffer from complexity in the form of complex tree with large number of hops and nodes.

The information gain was used as a splitting criteria of the modified ID3.

The proposed modified ID3 algorithm has been tested on a dataset for a different university employees with several attributes that directly affect their annual performance assessment.

The most optimized tree is constructed by taking one attribute that have the largest information gain from the dataset as a root of tree and repeating the process until the tree is completed.

The results showed that the proposed modified ID3 decision tree algorithm that based on Taneja entropy gives less complexity due to small tree with three nodes and two to one hope to reach the right decision.