IBM HR Analytics Employee Attrition & Performance

• Repository
• Kaggle Dataset

HR analytics is the process of collecting and analyzing Human Resource (HR) data in order to improve an organization’s workforce performance. The process can also be referred to as talent analytics, people analytics, or even workforce analytics. This method of data analysis takes data that is routinely collected by HR and correlates it to HR and organizational objectives. Doing so provides measured evidence of how HR initiatives are contributing to the organization’s goals and strategies.

Most organizations already have data that is routinely collected, so why the need for a specialized form of analytics? Can HR not simply look at the data they already have? Unfortunately, raw data on its own cannot actually provide any useful insight. It would be like looking at a large spreadsheet full of numbers and words. Once organized, compared and analyzed, this raw data provides useful insight. They can help answer questions like:

So lets start with our analysis!

Data Fields

This is a fictional data set created by IBM data scientists. This contain of 35 featured and 1470 row.

Import Package and Data

Started with imports of some basic libraries that are needed throughout the case. This includes Pandas and Numpy for data handling and processing as well as Matplotlib and Seaborn for visualization.

import numpy as np
import pandas as pd
import matplotlib.pyplot as plt
import seaborn as sns
import plotly.express as px

from sklearn.model_selection import train_test_split
from sklearn.linear_model import LogisticRegression


%matplotlib inline

For this exercise, the data set (.csv format) is downloaded to a local folder, read into the Jupyter notebook and stored in a Pandas DataFrame.

data = pd.read_csv('C:\My Files\Document\Coding\Datasheet\WA_Fn-UseC_-HR-Employee-Attrition.csv')
data.head()

Data Preparation

Get statistical information on numerical features and checking missing value.

Exploratory Data Analysis

How is attrition dependent on Age?

As seen in the chart above, the attrition is maximum between the age groups 28-32. The attrition rate keeps on falling with increasing age, as people look after stability in their jobs at these point of times. Also at a very younger age, i.e. from 18-20, the chances of an employee leaving the organization is far more- since they are exploring at that point of time. It reaches a break even point at the age of 21.

Is income the main factor towards employee attrition?

As seen in the above chart, the attrition rate is evidently high at very low income levels- less than 5k monthly. This decreases further- but a minor spike is noticed aorund 10k- indicating the middle class liveliood. They tend to shift towards a better standard of living, and hence move to a different job. When the monthly income is pretty decent, the chances of an employee leaving the organization is low- as seen by the flat line.

Does the Department of work impact attrition?

This data comprises of only 3 major departments- among which Sales department has the highest attrition rates (25.84%), followed by the Human Resource Department (19.05%). Research and Development has the least attrition rates, that suggests the stability and content of the department as can be seen from the chart above(13.83%).

How does the environment satisfaction impact attrition?

In the satisfaction Level 1-2, the chances of peope leaving the organization slightly decreases. This is indicative of the better hopes with which people stay in an organization. However, as we move from 2-3, people tend to move on to get better opportunities and experiences. The attrition rate is almost stagnant for the higher satisfaction levels.

How does self Job Satisfaction impact the Attrition?

With an increasing job satisfaction, the attrition rates decrease as can be seen in the chart above. Also from range 1-2 range we can infer (as seen above in Environment Satisfaction), the attrition level falls, but raises from 2-3, where the people tend to coose better opportunities.

Does company stocks for employees impact attrition?

The tendency of employees to leave the organization is much more when the stock availing options are limited. Since the stocks constitute to a huge amount of money while staying for a few years, people do not want to lose that opportunity. People with very limited/no stcok options have a freedom to leave the organization at will.

How does Work Life Balance impact the overall attrition rates?

People with poor levels of Work life balance have adjusted themselves to their jobs, but as seen for the above parameters with a better work life score, people are more accustomed to the better life and want to go for an attrition more. But this trend perishes when the work life balance is really good, and people are satisfied with the work they are doing.

How does work experience affect attrition?

As seen from the chart above, clearly, employees who started their career with the company- or have switched to the company in the initial years of their career, have a higher chances of leaving the organization to a different company. People who have gained much experience- working in multiple companies tend to stay in the company they join.

How does Work duration in current role impact Attrition?

We have seen people are more prone to leave the organization in the starting years on their role. When people are in the same role for a long period of time, they tend to stay longer for moving in an upward role.

Does Hike percentage impact Attrition?

Higher hikes motivate people to work better, and stay in the organization. Hence we see the chances of an employee leaving the organization where the hike is lower, is much more than a company that gives a good hike.

Are managers a reason of people resigning??

We notice 3 major spikes in the attrition rate, when we are analyzing the relationship of an employee with their manager. At the very start, where the time spent with the manager is relatively less- people tend to leave their jobs- considering their relationship with their previous managers. At an average span of 2 years, when employees feel they need an improvement, they also tend to go for a change. When the time spent with the manager is slightly higher (about 7 years)- people tend to find their career progression stagnant, and tend to go for a change. But when the relative time spend with a manager is very high- people are satisfied with their work. Hence the chances of an employee resigning then is significantly low.

Machine Learning Model

Lets try to predict for all the given inputs, how accurately can we we predict wether an employee will be staying in the organization or resigning from it with Logistic Regression

We got 84% accuracy, however, let us find method to improve this further and make the accuracy higher.

Conclusion

We have checked the data, and have come upon to infer the following observations:

1. People are tending to switch to a different jobs at the start of their careers, or at the earlier parts of it. Once they have settled with a family or have found stability in their jobs, they tend to stay long in the same organization- only going for vertical movements in the same organization.
2. Salary and stock ptions have a great motivation on the employees and people tend to leave the organization much lesser. Higher pay and more stock options have seen more employees remain loyal to their company.
3. Work life balance is a great motivation factor for the employees. However, people with a good work-life balance, tend to switch in search of better opportunities and a better standard of living.
4. Departments where target meeting performance is very much crucial (for e.g. Sales) tend to have a greater chances of leaving the organization as compared to departments with more administration perspective (For e.g. Human Resources)
5. People with a good Job Satisfaction and Environment satisfaction are loyal to the organization- and this speaks loud for any Organization. However, people who are not much satisfied with their current project- tend to leave the organization far more.