Two hidden layered based NN model with input features & target
Managing people is difficult than managing machines. Every year companies have to invest a lot in their employees and hence their proper management is a big concern for them. Taking the same into account, this project is about predicting whether an employee might leave the company in near future or not. The model predicts 'True' if the probablity of an employee leaving current position is greater than 0.5, or 'False' otherwise.
- How? Deep Learning (Neural Network) model is deployed for prediction
- Input: Age, DaileyRate, Department, DistanceFromHome, Education, JobInvolvement, JobLevel, MaritalStatus, PercentSalaryHike, YearsInCompany, YearsAtCurrentRole and 19 others
- Target: Classify employee based on his/her leave probablity
- Libraries: pandas, numpy, matplotlib, seaborn, sklearn, tensorflow
- Activation Function: ReLU & Sigmoid
- Optimization Function: Adam
- Insights:
- Average age of employee left (16.12%) is 33yrs. Older employee tend to stay longer.
- Employee doing over time, especially at low job level tend to leave early
- Higher trend of job change among unmarried employees
- About 40% of sales representative have already left
- Follow below link for complete step-by-step analysis
Github Link
Future Scope
Human behavior depends on many factor and predicting the same is a complex task. Accuracy depends on dataset and can vary but with few minor changes, the same model can be used for predicting any company's employee behaviour.