Enhancing Efficiency of Employment By Predicting Compensation Value of Applicants Team 5 John Liao, Jimmy Wang, Cesar Hsu, Jay Lim
Business problem Goal: To minimize cost of recruiting ● Missing potential talent Finding wrong person ● Clients: ● HR department of any firm + Headhunter companies
Data Mining Goal SALARY OUTPUT
Cutoff value About to 64,000 (USD)
Data Description ● Classification output: Suitable vs Unsuitable Input variable: 79 Problems 1. Cultural difference 2. Currency 3. Can not use PPP
4 Variable Selection 1 2 1. Logistic Regression 2. Gradient Boosted Tree 3. Collinearity Problem 4. Domain Knowledge 3 ● Reduce from 240 columns to 79 columns
Methods 1 2 What do Ensemble using logistic regression look like in RapidMiner?
Evaluation Ensemble Version with Oversampling Logistic Regression with Oversampling
Recommendations 1. Working culture in different industries is not considered 2. Some components of compensation can’t be evaluated in this model (such as basic salary, working key performance indicators, bonus, and welfare) 3. Renewing our model is needed after a period of time 4. The observations in our sample is small
Recommend
More recommend