A salary prediction system using machine learning techniques was developed by Dr. Fadi Sibai, Gulf University of Science & Technology Engineering Faculty and GEAR Center member, and his research colleagues at Wichita State University in the USA. A salary predictor has many uses. For instance, salary prediction helps employers and human resource organizations determine the appropriate salary to offer a new employee based on their profile. Likewise, employees can predict their potential salary based on their skills and years of experience in the job market, helping them in their decisions to seek other jobs, and accept or reject new job offers.
Experiments were done using the Kaggle dataset from the 1994 census database which has 32 thousand employee data records. The machine learning techniques used in determining whether an employee salary is less than or greater than $50,000 were: logistic regression, decision tree, Naive Bayes classifier, K-nearest neighbor, and support vector machine. Several attributes were considered in predicting the salary, including age, work class, final weight, education, marital status, occupation, relationship, race, gender, capital gain, capital loss, hours per week, native country, and salary. By training the machine learning models with this data, machine learning algorithms can identify patterns and relationships that contribute to salary levels, enabling accurate predictions to be made.

The methodology used is illustrated in Fig. 1. After importing the NumPy, Pandas, Matplotlib, and other libraries, and loading the dataset, ten different models were developed, and these models were trained with original training data and oversampled training data. In order to compare the models, Accuracy, as well as False Negative, and False Positive values were compared for the ten implemented models during the testing phase. Among the models evaluated, the Decision Tree with the original training data demonstrated superior performance in determining whether an employee’s salary is less than or greater than $50,000. In fact, the Decision Tree model exhibited a high accuracy rate of 82%. Consequently, this model becomes a valuable tool for employees, employers and HR Staff, and students seeking to make informed decisions about salaries.


Keywords:
decision tree; machine learning; model accuracy; salary prediction systems