SlideShare une entreprise Scribd logo
1  sur  44
IBM HR Analytics Employee
Attrition & Performance
Data Science Project
by: Angelin Grace Wijaya, Agarwala Pratham, Krishna Shivangi
SOURCE
- Kaggle Dataset by pavansubhash
- Fictional dataset created by IBM data scientists
- Contains 35 columns and 1470 records
- No missing values
https://www.kaggle.com/pavansubhasht/ibm-hr-analytics-attrition-dataset
Business Objective of our Project
- HR Analysis tool for companies especially those with
large workforce.
- Help companies to be prepared for future
employee-loss
- To find possible reasons for employee attrition, in
order to prevent valuable employees from leaving.
- Predicting employee’s Performance Rating and hence
distributing the employees into two classes
( 0 : Low Performance Rating, 1 : High Performance Rating)
- Accordingly, predict the employees “Percent Salary
Hike” for the next working term.
DATA CLEANING
Two of the columns in the employee dataset are labelled:
“Standard Hours” and “Employee Count”.
In “Standard Hours”, we found that all the records have a value of 80. What does this mean?
This standard hour is also called a 9/80 work schedule. Here, employees work for 80 hours over a period
of 9 days.
In “Employee Count”, we found that all the records have a value of 1. What does this mean?
Employee count is also known as headcount. It is the number of employees working in the company,
where each employee, whether full-time or part-time, is counted as 1.
As these columns all have the same values, we decided to drop them while keeping in mind of what each
of them represents.
DATA CLEANING
Two of the columns in the employee dataset are labelled:
“Over18” and “NumCompaniesWorked”.
In “Over18”, we found that all the records have a value of “Yes”.
We found that there are actually 8 employees who are 18 from the “Age” column. This means that the “Over18”
column has mismatched values, or that this column actually means employees who are 18 and above.
We kept this column and change the records for those who are aged 18 to “No” in their “Over 18” column.
In “NumCompaniesWorked”, we found that there are records with zero companies worked.
We found that for these employees, they all have 1 year where they worked not for a company (ex. part-time).
As 18 year-old employees, who have zero years of work experience, have one as number of companies worked (their
current one), we also decided to change the records of number of companies worked as zero to one.
Exploratory Data Analysis
JOBS
Sales department:
- Sales Executive
- Sales Representative
- Sales Manager
Research & Development department:
- Research Scientist
- Laboratory Technician
- Manufacturing Director
- Healthcare Representative
- Research Director
- Research Manager
Human Resources department:
- Human Resources
- Human Resources Manager
JOB LEVELS
It is defined as the classification of
authority and/or salary levels of positions
in an organization.
Ex. senior executives, executive, middle management,
advisors, associate
Different job levels have different monthly income.
JOB ROLE & AGE
- Managers and Research Directors
have the oldest median age.
- Managers and Research Directors
are also earning the highest
median monthly income
- Sales representatives are the
youngest and they are receiving
the lowest monthly income
JOB ROLE WITH TOTAL WORKING YEARS & YEARS IN CURRENT ROLE
- Managers and research directors have the highest
total working years and years at current company
- Sales representatives have the lowest total working
years and years at current company
- Years since last promotion is more or less the same
for all the job roles other than managers
Predicting Attrition of
Employees
WHAT IS ATTRITION?
Attrition is the decrease in the number of employees over time as they leave due to personal
reasons or retire.
Low attrition rate is desired by many companies because it implies that employees are satisfied
with their workplace and the company does not have to train new recruits.
Disadvantages of Attrition:
1. Decrease Overall Performance
2. Difficulty in Managing Task (that was left by the employee)
3. Increased Cost (must hire and train new employee)
4. Lack of Knowledgeable/Experienced Employee
5. Losing Out on Employee Development Plan
6. Negative Image
DECISION TREE
We found that our dataset is
imbalanced.
The ratio of the number of people with
attrition to without attrition is 1233 :
237
The confusion matrix above is before
balancing.
Those who actually have attrition are
being classified as having no attrition.
Below is the matrix after balancing.
Accuracy Train: 0.86 Accuracy Test: 0.83
Accuracy Train: 0.74 Accuracy Test: 0.77
RANDOM FOREST
We used GridSearchCV to find the ideal number of trees
and depth for our random forest.
Result: {'max_depth': 7, 'n_estimators': 100}
Below is our confusion matrix for random forest:
Accuracy Train: 0.95 Accuracy Test: 0.90
Cross-validation is performed by the
GridsearchCV.
We pass the whole dataset to it
without splitting it and by default it
performs 5-fold cross validation
Why random forest?
A single decision tree like the one in
the previous slide brings a much
lower accuracy than forest.
In random forest, we can plot
multiple trees that learn from one
another’s mistake (or wrong
prediction) to give a better accuracy.
ROC CURVES
We also found the Best Threshold to be 0.498017 and optimal G-Mean as 0.951
VARIABLE IMPORTANCE
1. Overtime
2. Monthly Income
3. Years At Company
7. Total Working Years
8. Stock Option Level
9. Years With Current Manager
4. Job Level
5. Age
6. Years In Current Role
Train Test Split has been fixed with random state of 1.
Without random state, these 9 variables are still in the top 9 importance but may be in different ranks.
(ex. Monthly Income may become more important than Overtime, but is still within the 9 ranks)
Attrition and
Monthly Income
Employees with attrition have lower
monthly income
Attrition also has a -0.159840 correlation with
Monthly Income
MONTHLY INCOME- MORE INSIGHTS
First of all, Monthly Income is a continuous data, so we performed extensive EDA to study the patterns in
Monthly Income .
1. RFECV : Recursive Feature Elimination from sklearn.feature_selection an for CV we will will use
KFold validation.
2. Greedy Search : Iteratively removing variables from the dataset until the there are only five
variables left excluding the response variable(MonthlyIncome) and criterion is the accuracy score of
the model . It removes the variables that least increases the accuracy score .
MONTHLY INCOME - RFECV
Using RFECV(Recursive Feature Elimination CVross Validation) from sklearn.feature_slection we performed the following the
steps:
1. We label-encoded the variables which have some kind of order or have classes 2 or less, and for rest we did One-Hot
Encoding.
2. Removing correlated features is done because we don’t want highly correlated features in our dataset as they provide the
same information — one is enough. For this step we wrote a simple algorithm. We iterated through the entire correlation
matrix and removed variables with correlation with other variables exceeding 0.75.
3. We declared two variables — X and target where first represents all the features, and the second represents the target
variable. Then we’ll use an Machine learning Model (We are using LinearRegression). Then we ran RFECV. Here are the
arguments:
a. Estimator : LinearRegression()
b. Cv : KFold(10)
c. Scoring : ‘ Accuracy’
4. Now seeing the graph the optimal of feature is 8 to get
Best accuracy for LinearRegression Model.
➔ TotalWorkingYears
➔ Attrition
➔ Department
➔ JobRole
➔ Over18
➔ BusinessTravel_Non-Travel
➔ EducationField_Human
Resources
➔ 'EducationField_Marketing
Variables Short-listed after using RFECV:
Attrition and
Years (1)
Employees with attrition have lower
years with current manager and total
working years
Attrition has a -0.156199 correlation with Years With Current
Manager and a -0.171063 correlation with Total Working
Years
Age and Total working years:
● The age has been classified as this:
- 20s → younger employees
- 40s → middle aged employees
- 50s → older employees
● The correlation between total working years
and age is 0.68
● For all three categories, the number of
employees who have worked for a shorter
period of time are the ones who want to attire
● As the age increases, the total number of
working years increases
Attrition and
Years (2)
Employees with attrition have lower
years at company and lower years in
current role
Attrition has a -0.134392 correlation with Years At Company
and a -0.160545 correlation with Years In Current Role
Among attiring employees,
- Sales Manager and Research Director are attiring after longer years at company.
- We know that Managers and Research Directors have the oldest median age and the highest median monthly income. Therefore, it is highly
possible that they are attiring due to retirement.
- Employees that attire with the least number of working years are sales representatives.
- Other job roles are attiring after shorter working years, and we predict that it is to look for better job opportunities.
Attrition and
Job Level
Employees with a lower job level
are more likely to attire
Attrition has a -0.169105 correlation with Job Level
Job level and Monthly Income
- The correlation between job level and monthly
income is 0.95
- Employees with a job level of 5 are either
managers and research directors (highest paid)
- Employees with a job level of 1 are sales
representatives (lowest paid)
- As the job level increases, the number of
employees that want to attire decrease
- Job level of 1 has the highest number of
employees that want to attire
From this boxplot, we can see how every job level almost has a corresponding range of monthly income.
The difference in median monthly income (No Attrition - Attrition ) is greatest for Job Level= 4 followed by Job Level= 1.
A possible reason for high attrition rate at Job Level 1 may be that they feel it is unfair to be receiving a lower monthly income
than others in the same job level.
How about Job level 2 then? The monthly income for both are almost equal, so what could be the reason for them attiring?
Also, the difference in monthly income in level 4 is very high, why is the attrition rate lower?
Possible reason for high attrition in Job Level 2:
- Even though they have nearly the same
median monthly income, it turns out that
employees here with attrition have a lower
median stock option level compared to
those not attiring.
Possible reason for low attrition in Job Level 4:
- Even though there is a huge gap in median
monthly income, most of the employees
have a high monthly income that may
compensate for a lower stock option level.
The red, box for those very few employees
with attrition have no stocks and a much
low monthly income compared to the rest.
Attrition and
Age
There are more younger people
with attrition
Attrition has a -0.159205 correlation with Age
Age was split into 3 categories:
● Younger → aged between 18 and 30, inclusive. There is a 0.259 probability of attiring
● Middle → aged between 31 and 42 inclusive. There is a 0.132 probability of attiring
● Older → aged between 43 and 60 inclusive. There is a 0.116 probability of attiring
This grouping aided us in understanding from which age group are most attired employees found.
These are the graphs of age against attrition:
younger middle older
There is a higher probability for the younger group aged 18 to 30 to attire than other age
groups. Employees may attire to look for better job opportunities outside the company
while they are young.
Attrition and
Overtime
Employees with attrition are
more likely those that do not
have overtime
Attrition has a 0.246118 correlation with Overtime
Employees without attrition, whether working overtime or not have a median job satisfaction at 3.
However, attiring employees not working overtime are actually less satisfied with their job than those
working overtime.
median = 3 median = 2 median = 3
Q1= 2
Q1= 1
Q1= 2
Q3= 4
In this box plot, we can see how job level 1 employees with attrition and no overtime have a much lower median monthly
income for every job involvement value.
This could mean that attiring employees feel unfair because they are being paid less than people in the same job level who
are also not working overtime and with the same job involvement.
This trend is observed for all the other job levels in our Jupyter notebook
Attrition and
Stock Option
Level
Employees with attrition are
most likely those without any
stock option
Attrition has a -0.137145 correlation with Stock Option
Level
For every job role except Human Resources, there seems to be a trend where employees with
attrition are those that have a lower stock level compared to those without attrition.
Employees might be attiring because they have a lower stock option level than other
employees of the same role.
PUTTING IT ALL TOGETHER
We used Plotly to
produce this density
heatmap.
Most employees with
attrition have a low Job
Level, low Years at
Company, low Stock
Option level, and low
Monthly Income
Performance Rating of
Employees
WHAT IS PERFORMANCE RATING? WHAT ARE WE PLANNING TO DO WITH IT?
- Performance Rating is a multi-class variable which
partitions the employees into 4 groups.
a. Performance Rating of 1
b. Performance Rating of 2
c. Performance Rating of 3
d. Performance Rating of 4
- However we found that the employees only have either a
performance rating of 3 or 4
- Thus, we took performance rating as a binary-class
variable
- Our assumption is supported by the adjacent graph.
Which clearly shows, company provides higher increase
in salary to employees with Performance Rating of 4.
- We will use the predicted values of PerformanceRating to
predict the values of PercentSalaryHike of respective
employees.
DECISION TREE
Performance Rating was highly imbalanced so we
upsampled the data using “upsample” from sklearn.utils
After balancing the data we ran GridSearchCV for tuning
max_depth (found it to be 10) and
Before Balancing
Accuracy: 0.85
After Balancing Train After Balancing Test
Accuracy: 0.90 Accuracy: 0.79
RANDOM FOREST & VARIABLE IMPORTANCE
We used GridSearchCV to find the ideal number of trees and depth for our
random forest.
RandomForestClassifier(max_depth=9, n_estimators=150)
After Balancing Test
Accuracy: 0.95
PREDICTING PERCENT SALARY HIKE USING
PREDICTED VALUES OF PERFORMANCE RATING
We know that PercentSalaryHike and PerformanceRating are perfectly
correlated in terms of classification, whereas linearly has a correlation
of 0.77.
So, we can treat PercentSalaryHike in two ways :
● As a continuous variable and perform a regression model.
○ RandomForestRegressor
● As a multi-class variable and perform a multi-classification
model.
○ RandomForestClassifier
RandomForestClassifier
RandomForestRegressor RandomForestClassifier
Accuracy Test: 0.841 Classification Accuracy Test : 0.55
So comparing the accuracy of the models we decided to go with the regression model to predict
PercentSalaryHike
“Research indicates that workers have three prime needs:
Interesting work, recognition for doing a good job, and
being let in on things that are going on in the company.”
- Zig Ziglar
CONCLUSION
To have a low attrition rate, it is important for companies to:
1. Fairly pay employees in same job level, same job involvement, same job role
with almost equal monthly income and stocks.
2. Offer stock options to more employees especially new recruits as an
incentive.
3. Increase allowance for employees to work overtime, and make sure that
those working overtime are being paid more than those who are not.
4. Also from our quote, companies should remember to appreciate their
employees and work on the harmonic balance of their company.
THANK YOU
From: DS2 Group 7

Contenu connexe

Tendances

Employee retention issues and analysis
Employee retention issues and analysisEmployee retention issues and analysis
Employee retention issues and analysisRehan Akhtar
 
Hr analytics project
Hr analytics projectHr analytics project
Hr analytics projectJatin Saini
 
Employee Attrition
Employee AttritionEmployee Attrition
Employee AttritionVinay sattur
 
Predictive HR--Analytics
Predictive HR--AnalyticsPredictive HR--Analytics
Predictive HR--AnalyticsE J Sarma
 
HR Analytics: Using Machine Learning to Predict Employee Turnover - Matt Danc...
HR Analytics: Using Machine Learning to Predict Employee Turnover - Matt Danc...HR Analytics: Using Machine Learning to Predict Employee Turnover - Matt Danc...
HR Analytics: Using Machine Learning to Predict Employee Turnover - Matt Danc...Sri Ambati
 
The strategic Role of HRM
The strategic Role of HRMThe strategic Role of HRM
The strategic Role of HRMQamar Farooq
 
Machine Learning Approach for Employee Attrition Analysis
Machine Learning Approach for Employee Attrition AnalysisMachine Learning Approach for Employee Attrition Analysis
Machine Learning Approach for Employee Attrition Analysisijtsrd
 
Predictive Analytics for HR: A Primer to Get Started on your HR Analytics Jou...
Predictive Analytics for HR: A Primer to Get Started on your HR Analytics Jou...Predictive Analytics for HR: A Primer to Get Started on your HR Analytics Jou...
Predictive Analytics for HR: A Primer to Get Started on your HR Analytics Jou...Dr Susan Entwisle
 
Recruiting Metrics Presentation
Recruiting Metrics PresentationRecruiting Metrics Presentation
Recruiting Metrics PresentationRachel Webb
 
Managing attrition rate in BPO's and KPO's
Managing attrition rate in BPO's and KPO'sManaging attrition rate in BPO's and KPO's
Managing attrition rate in BPO's and KPO'sMadhuri Lotankar
 
Presentation on HR Planning
Presentation on HR PlanningPresentation on HR Planning
Presentation on HR PlanningKamlesh Thakur
 

Tendances (20)

Employee retention issues and analysis
Employee retention issues and analysisEmployee retention issues and analysis
Employee retention issues and analysis
 
Hr analytics project
Hr analytics projectHr analytics project
Hr analytics project
 
Employee Attrition
Employee AttritionEmployee Attrition
Employee Attrition
 
Hr analytics
Hr analyticsHr analytics
Hr analytics
 
Predictive HR--Analytics
Predictive HR--AnalyticsPredictive HR--Analytics
Predictive HR--Analytics
 
attrition analysis
attrition analysisattrition analysis
attrition analysis
 
HR Analytics: Using Machine Learning to Predict Employee Turnover - Matt Danc...
HR Analytics: Using Machine Learning to Predict Employee Turnover - Matt Danc...HR Analytics: Using Machine Learning to Predict Employee Turnover - Matt Danc...
HR Analytics: Using Machine Learning to Predict Employee Turnover - Matt Danc...
 
HR Analytics
HR AnalyticsHR Analytics
HR Analytics
 
Employee Attrition Analysis
Employee Attrition AnalysisEmployee Attrition Analysis
Employee Attrition Analysis
 
Hr analytics
Hr analyticsHr analytics
Hr analytics
 
People Analytics.pptx
People Analytics.pptxPeople Analytics.pptx
People Analytics.pptx
 
The strategic Role of HRM
The strategic Role of HRMThe strategic Role of HRM
The strategic Role of HRM
 
Presentation on Attrition
 Presentation on Attrition Presentation on Attrition
Presentation on Attrition
 
Machine Learning Approach for Employee Attrition Analysis
Machine Learning Approach for Employee Attrition AnalysisMachine Learning Approach for Employee Attrition Analysis
Machine Learning Approach for Employee Attrition Analysis
 
HR ANALYTICS
HR ANALYTICS HR ANALYTICS
HR ANALYTICS
 
Predictive Analytics for HR: A Primer to Get Started on your HR Analytics Jou...
Predictive Analytics for HR: A Primer to Get Started on your HR Analytics Jou...Predictive Analytics for HR: A Primer to Get Started on your HR Analytics Jou...
Predictive Analytics for HR: A Primer to Get Started on your HR Analytics Jou...
 
Recruiting Metrics Presentation
Recruiting Metrics PresentationRecruiting Metrics Presentation
Recruiting Metrics Presentation
 
Analytics in hr
Analytics in hrAnalytics in hr
Analytics in hr
 
Managing attrition rate in BPO's and KPO's
Managing attrition rate in BPO's and KPO'sManaging attrition rate in BPO's and KPO's
Managing attrition rate in BPO's and KPO's
 
Presentation on HR Planning
Presentation on HR PlanningPresentation on HR Planning
Presentation on HR Planning
 

Similaire à IBM HR Analytics Employee Attrition & Performance

Gupta ayankprojectassignmnet
Gupta ayankprojectassignmnetGupta ayankprojectassignmnet
Gupta ayankprojectassignmnetAyank Gupta
 
6 Cutting-Edge HR Metrics to Measure in 2019
6 Cutting-Edge HR Metrics to Measure in 20196 Cutting-Edge HR Metrics to Measure in 2019
6 Cutting-Edge HR Metrics to Measure in 2019Namely
 
Employee Retension Capstone Project - Neeraj Bubby.pptx
Employee Retension Capstone Project - Neeraj Bubby.pptxEmployee Retension Capstone Project - Neeraj Bubby.pptx
Employee Retension Capstone Project - Neeraj Bubby.pptxBoston Institute of Analytics
 
46834Strategic Human Resource Management - Session 6.pptx
46834Strategic Human Resource Management - Session 6.pptx46834Strategic Human Resource Management - Session 6.pptx
46834Strategic Human Resource Management - Session 6.pptxMuhammadAsghar955025
 
HRM Case STUDY Qusetion AND Answer
HRM Case STUDY Qusetion AND AnswerHRM Case STUDY Qusetion AND Answer
HRM Case STUDY Qusetion AND AnswerMuhammad Manan
 
Database Marketing Project Slides
Database Marketing Project SlidesDatabase Marketing Project Slides
Database Marketing Project SlidesIshanMalpotra
 
Salary survey c level-2018
Salary survey c level-2018Salary survey c level-2018
Salary survey c level-2018Olga Novykova
 
Salary survey c level-2021
Salary survey c level-2021Salary survey c level-2021
Salary survey c level-2021Kristina Florya
 
9 Salary Surveys A Snapshot Market data obtained from salary surv.docx
9 Salary Surveys A Snapshot Market data obtained from salary surv.docx9 Salary Surveys A Snapshot Market data obtained from salary surv.docx
9 Salary Surveys A Snapshot Market data obtained from salary surv.docxsleeperharwell
 
How to retain and motivate Staff DaNang December 2013
How to retain and motivate Staff DaNang December 2013How to retain and motivate Staff DaNang December 2013
How to retain and motivate Staff DaNang December 2013Tom Vovers
 
www.tex-cetera.com CHAPTER 9 Compe.docx
www.tex-cetera.com CHAPTER 9 Compe.docxwww.tex-cetera.com CHAPTER 9 Compe.docx
www.tex-cetera.com CHAPTER 9 Compe.docxericbrooks84875
 
Professional Performance Metrics | A Framework to Optimize decision Making | ...
Professional Performance Metrics | A Framework to Optimize decision Making | ...Professional Performance Metrics | A Framework to Optimize decision Making | ...
Professional Performance Metrics | A Framework to Optimize decision Making | ...Muhammad Nizam Uddin
 
Assessing and analyzing the effectiveness of recruiting
Assessing and analyzing the effectiveness of recruitingAssessing and analyzing the effectiveness of recruiting
Assessing and analyzing the effectiveness of recruitingShubham Singhal
 
HUMAN RESOURCE ANALYTICS METRICS
HUMAN RESOURCE ANALYTICS METRICSHUMAN RESOURCE ANALYTICS METRICS
HUMAN RESOURCE ANALYTICS METRICSHoney Ramchandani
 
Performance evaluation sample forms
Performance evaluation sample formsPerformance evaluation sample forms
Performance evaluation sample formsgarethbale741
 
The Next Generation Of Employee Referral Programs Will Focus On Improving Qua...
The Next Generation Of Employee Referral Programs Will Focus On Improving Qua...The Next Generation Of Employee Referral Programs Will Focus On Improving Qua...
The Next Generation Of Employee Referral Programs Will Focus On Improving Qua...Dr. John Sullivan
 
Metrics that your ceo cares about
Metrics that your ceo cares aboutMetrics that your ceo cares about
Metrics that your ceo cares aboutDr. John Sullivan
 

Similaire à IBM HR Analytics Employee Attrition & Performance (20)

Gupta ayankprojectassignmnet
Gupta ayankprojectassignmnetGupta ayankprojectassignmnet
Gupta ayankprojectassignmnet
 
6 Cutting-Edge HR Metrics to Measure in 2019
6 Cutting-Edge HR Metrics to Measure in 20196 Cutting-Edge HR Metrics to Measure in 2019
6 Cutting-Edge HR Metrics to Measure in 2019
 
Employee Retension Capstone Project - Neeraj Bubby.pptx
Employee Retension Capstone Project - Neeraj Bubby.pptxEmployee Retension Capstone Project - Neeraj Bubby.pptx
Employee Retension Capstone Project - Neeraj Bubby.pptx
 
46834Strategic Human Resource Management - Session 6.pptx
46834Strategic Human Resource Management - Session 6.pptx46834Strategic Human Resource Management - Session 6.pptx
46834Strategic Human Resource Management - Session 6.pptx
 
HRM Case STUDY Qusetion AND Answer
HRM Case STUDY Qusetion AND AnswerHRM Case STUDY Qusetion AND Answer
HRM Case STUDY Qusetion AND Answer
 
Database Marketing Project Slides
Database Marketing Project SlidesDatabase Marketing Project Slides
Database Marketing Project Slides
 
HR Data Analysis.pdf
HR Data Analysis.pdfHR Data Analysis.pdf
HR Data Analysis.pdf
 
Salary survey c level-2018
Salary survey c level-2018Salary survey c level-2018
Salary survey c level-2018
 
Salary survey c level-2021
Salary survey c level-2021Salary survey c level-2021
Salary survey c level-2021
 
Why SAP BASIS?
Why SAP BASIS?Why SAP BASIS?
Why SAP BASIS?
 
Salary survey С-level 2019
Salary survey С-level 2019Salary survey С-level 2019
Salary survey С-level 2019
 
9 Salary Surveys A Snapshot Market data obtained from salary surv.docx
9 Salary Surveys A Snapshot Market data obtained from salary surv.docx9 Salary Surveys A Snapshot Market data obtained from salary surv.docx
9 Salary Surveys A Snapshot Market data obtained from salary surv.docx
 
How to retain and motivate Staff DaNang December 2013
How to retain and motivate Staff DaNang December 2013How to retain and motivate Staff DaNang December 2013
How to retain and motivate Staff DaNang December 2013
 
www.tex-cetera.com CHAPTER 9 Compe.docx
www.tex-cetera.com CHAPTER 9 Compe.docxwww.tex-cetera.com CHAPTER 9 Compe.docx
www.tex-cetera.com CHAPTER 9 Compe.docx
 
Professional Performance Metrics | A Framework to Optimize decision Making | ...
Professional Performance Metrics | A Framework to Optimize decision Making | ...Professional Performance Metrics | A Framework to Optimize decision Making | ...
Professional Performance Metrics | A Framework to Optimize decision Making | ...
 
Assessing and analyzing the effectiveness of recruiting
Assessing and analyzing the effectiveness of recruitingAssessing and analyzing the effectiveness of recruiting
Assessing and analyzing the effectiveness of recruiting
 
HUMAN RESOURCE ANALYTICS METRICS
HUMAN RESOURCE ANALYTICS METRICSHUMAN RESOURCE ANALYTICS METRICS
HUMAN RESOURCE ANALYTICS METRICS
 
Performance evaluation sample forms
Performance evaluation sample formsPerformance evaluation sample forms
Performance evaluation sample forms
 
The Next Generation Of Employee Referral Programs Will Focus On Improving Qua...
The Next Generation Of Employee Referral Programs Will Focus On Improving Qua...The Next Generation Of Employee Referral Programs Will Focus On Improving Qua...
The Next Generation Of Employee Referral Programs Will Focus On Improving Qua...
 
Metrics that your ceo cares about
Metrics that your ceo cares aboutMetrics that your ceo cares about
Metrics that your ceo cares about
 

Dernier

CebaBaby dropshipping via API with DroFX.pptx
CebaBaby dropshipping via API with DroFX.pptxCebaBaby dropshipping via API with DroFX.pptx
CebaBaby dropshipping via API with DroFX.pptxolyaivanovalion
 
Call Girls Bannerghatta Road Just Call 👗 7737669865 👗 Top Class Call Girl Ser...
Call Girls Bannerghatta Road Just Call 👗 7737669865 👗 Top Class Call Girl Ser...Call Girls Bannerghatta Road Just Call 👗 7737669865 👗 Top Class Call Girl Ser...
Call Girls Bannerghatta Road Just Call 👗 7737669865 👗 Top Class Call Girl Ser...amitlee9823
 
ELKO dropshipping via API with DroFx.pptx
ELKO dropshipping via API with DroFx.pptxELKO dropshipping via API with DroFx.pptx
ELKO dropshipping via API with DroFx.pptxolyaivanovalion
 
Call Girls Jalahalli Just Call 👗 7737669865 👗 Top Class Call Girl Service Ban...
Call Girls Jalahalli Just Call 👗 7737669865 👗 Top Class Call Girl Service Ban...Call Girls Jalahalli Just Call 👗 7737669865 👗 Top Class Call Girl Service Ban...
Call Girls Jalahalli Just Call 👗 7737669865 👗 Top Class Call Girl Service Ban...amitlee9823
 
Generative AI on Enterprise Cloud with NiFi and Milvus
Generative AI on Enterprise Cloud with NiFi and MilvusGenerative AI on Enterprise Cloud with NiFi and Milvus
Generative AI on Enterprise Cloud with NiFi and MilvusTimothy Spann
 
Accredited-Transport-Cooperatives-Jan-2021-Web.pdf
Accredited-Transport-Cooperatives-Jan-2021-Web.pdfAccredited-Transport-Cooperatives-Jan-2021-Web.pdf
Accredited-Transport-Cooperatives-Jan-2021-Web.pdfadriantubila
 
Call Girls Bommasandra Just Call 👗 7737669865 👗 Top Class Call Girl Service B...
Call Girls Bommasandra Just Call 👗 7737669865 👗 Top Class Call Girl Service B...Call Girls Bommasandra Just Call 👗 7737669865 👗 Top Class Call Girl Service B...
Call Girls Bommasandra Just Call 👗 7737669865 👗 Top Class Call Girl Service B...amitlee9823
 
Escorts Service Kumaraswamy Layout ☎ 7737669865☎ Book Your One night Stand (B...
Escorts Service Kumaraswamy Layout ☎ 7737669865☎ Book Your One night Stand (B...Escorts Service Kumaraswamy Layout ☎ 7737669865☎ Book Your One night Stand (B...
Escorts Service Kumaraswamy Layout ☎ 7737669865☎ Book Your One night Stand (B...amitlee9823
 
Mg Road Call Girls Service: 🍓 7737669865 🍓 High Profile Model Escorts | Banga...
Mg Road Call Girls Service: 🍓 7737669865 🍓 High Profile Model Escorts | Banga...Mg Road Call Girls Service: 🍓 7737669865 🍓 High Profile Model Escorts | Banga...
Mg Road Call Girls Service: 🍓 7737669865 🍓 High Profile Model Escorts | Banga...amitlee9823
 
Call Girls Indiranagar Just Call 👗 7737669865 👗 Top Class Call Girl Service B...
Call Girls Indiranagar Just Call 👗 7737669865 👗 Top Class Call Girl Service B...Call Girls Indiranagar Just Call 👗 7737669865 👗 Top Class Call Girl Service B...
Call Girls Indiranagar Just Call 👗 7737669865 👗 Top Class Call Girl Service B...amitlee9823
 
Capstone Project on IBM Data Analytics Program
Capstone Project on IBM Data Analytics ProgramCapstone Project on IBM Data Analytics Program
Capstone Project on IBM Data Analytics ProgramMoniSankarHazra
 
Al Barsha Escorts $#$ O565212860 $#$ Escort Service In Al Barsha
Al Barsha Escorts $#$ O565212860 $#$ Escort Service In Al BarshaAl Barsha Escorts $#$ O565212860 $#$ Escort Service In Al Barsha
Al Barsha Escorts $#$ O565212860 $#$ Escort Service In Al BarshaAroojKhan71
 
Cheap Rate Call girls Sarita Vihar Delhi 9205541914 shot 1500 night
Cheap Rate Call girls Sarita Vihar Delhi 9205541914 shot 1500 nightCheap Rate Call girls Sarita Vihar Delhi 9205541914 shot 1500 night
Cheap Rate Call girls Sarita Vihar Delhi 9205541914 shot 1500 nightDelhi Call girls
 
Digital Advertising Lecture for Advanced Digital & Social Media Strategy at U...
Digital Advertising Lecture for Advanced Digital & Social Media Strategy at U...Digital Advertising Lecture for Advanced Digital & Social Media Strategy at U...
Digital Advertising Lecture for Advanced Digital & Social Media Strategy at U...Valters Lauzums
 
Invezz.com - Grow your wealth with trading signals
Invezz.com - Grow your wealth with trading signalsInvezz.com - Grow your wealth with trading signals
Invezz.com - Grow your wealth with trading signalsInvezz1
 
BigBuy dropshipping via API with DroFx.pptx
BigBuy dropshipping via API with DroFx.pptxBigBuy dropshipping via API with DroFx.pptx
BigBuy dropshipping via API with DroFx.pptxolyaivanovalion
 

Dernier (20)

CebaBaby dropshipping via API with DroFX.pptx
CebaBaby dropshipping via API with DroFX.pptxCebaBaby dropshipping via API with DroFX.pptx
CebaBaby dropshipping via API with DroFX.pptx
 
Predicting Loan Approval: A Data Science Project
Predicting Loan Approval: A Data Science ProjectPredicting Loan Approval: A Data Science Project
Predicting Loan Approval: A Data Science Project
 
Call Girls Bannerghatta Road Just Call 👗 7737669865 👗 Top Class Call Girl Ser...
Call Girls Bannerghatta Road Just Call 👗 7737669865 👗 Top Class Call Girl Ser...Call Girls Bannerghatta Road Just Call 👗 7737669865 👗 Top Class Call Girl Ser...
Call Girls Bannerghatta Road Just Call 👗 7737669865 👗 Top Class Call Girl Ser...
 
Abortion pills in Doha Qatar (+966572737505 ! Get Cytotec
Abortion pills in Doha Qatar (+966572737505 ! Get CytotecAbortion pills in Doha Qatar (+966572737505 ! Get Cytotec
Abortion pills in Doha Qatar (+966572737505 ! Get Cytotec
 
CHEAP Call Girls in Saket (-DELHI )🔝 9953056974🔝(=)/CALL GIRLS SERVICE
CHEAP Call Girls in Saket (-DELHI )🔝 9953056974🔝(=)/CALL GIRLS SERVICECHEAP Call Girls in Saket (-DELHI )🔝 9953056974🔝(=)/CALL GIRLS SERVICE
CHEAP Call Girls in Saket (-DELHI )🔝 9953056974🔝(=)/CALL GIRLS SERVICE
 
ELKO dropshipping via API with DroFx.pptx
ELKO dropshipping via API with DroFx.pptxELKO dropshipping via API with DroFx.pptx
ELKO dropshipping via API with DroFx.pptx
 
Call Girls Jalahalli Just Call 👗 7737669865 👗 Top Class Call Girl Service Ban...
Call Girls Jalahalli Just Call 👗 7737669865 👗 Top Class Call Girl Service Ban...Call Girls Jalahalli Just Call 👗 7737669865 👗 Top Class Call Girl Service Ban...
Call Girls Jalahalli Just Call 👗 7737669865 👗 Top Class Call Girl Service Ban...
 
Generative AI on Enterprise Cloud with NiFi and Milvus
Generative AI on Enterprise Cloud with NiFi and MilvusGenerative AI on Enterprise Cloud with NiFi and Milvus
Generative AI on Enterprise Cloud with NiFi and Milvus
 
Accredited-Transport-Cooperatives-Jan-2021-Web.pdf
Accredited-Transport-Cooperatives-Jan-2021-Web.pdfAccredited-Transport-Cooperatives-Jan-2021-Web.pdf
Accredited-Transport-Cooperatives-Jan-2021-Web.pdf
 
Call Girls Bommasandra Just Call 👗 7737669865 👗 Top Class Call Girl Service B...
Call Girls Bommasandra Just Call 👗 7737669865 👗 Top Class Call Girl Service B...Call Girls Bommasandra Just Call 👗 7737669865 👗 Top Class Call Girl Service B...
Call Girls Bommasandra Just Call 👗 7737669865 👗 Top Class Call Girl Service B...
 
Escorts Service Kumaraswamy Layout ☎ 7737669865☎ Book Your One night Stand (B...
Escorts Service Kumaraswamy Layout ☎ 7737669865☎ Book Your One night Stand (B...Escorts Service Kumaraswamy Layout ☎ 7737669865☎ Book Your One night Stand (B...
Escorts Service Kumaraswamy Layout ☎ 7737669865☎ Book Your One night Stand (B...
 
Call Girls In Shalimar Bagh ( Delhi) 9953330565 Escorts Service
Call Girls In Shalimar Bagh ( Delhi) 9953330565 Escorts ServiceCall Girls In Shalimar Bagh ( Delhi) 9953330565 Escorts Service
Call Girls In Shalimar Bagh ( Delhi) 9953330565 Escorts Service
 
Mg Road Call Girls Service: 🍓 7737669865 🍓 High Profile Model Escorts | Banga...
Mg Road Call Girls Service: 🍓 7737669865 🍓 High Profile Model Escorts | Banga...Mg Road Call Girls Service: 🍓 7737669865 🍓 High Profile Model Escorts | Banga...
Mg Road Call Girls Service: 🍓 7737669865 🍓 High Profile Model Escorts | Banga...
 
Call Girls Indiranagar Just Call 👗 7737669865 👗 Top Class Call Girl Service B...
Call Girls Indiranagar Just Call 👗 7737669865 👗 Top Class Call Girl Service B...Call Girls Indiranagar Just Call 👗 7737669865 👗 Top Class Call Girl Service B...
Call Girls Indiranagar Just Call 👗 7737669865 👗 Top Class Call Girl Service B...
 
Capstone Project on IBM Data Analytics Program
Capstone Project on IBM Data Analytics ProgramCapstone Project on IBM Data Analytics Program
Capstone Project on IBM Data Analytics Program
 
Al Barsha Escorts $#$ O565212860 $#$ Escort Service In Al Barsha
Al Barsha Escorts $#$ O565212860 $#$ Escort Service In Al BarshaAl Barsha Escorts $#$ O565212860 $#$ Escort Service In Al Barsha
Al Barsha Escorts $#$ O565212860 $#$ Escort Service In Al Barsha
 
Cheap Rate Call girls Sarita Vihar Delhi 9205541914 shot 1500 night
Cheap Rate Call girls Sarita Vihar Delhi 9205541914 shot 1500 nightCheap Rate Call girls Sarita Vihar Delhi 9205541914 shot 1500 night
Cheap Rate Call girls Sarita Vihar Delhi 9205541914 shot 1500 night
 
Digital Advertising Lecture for Advanced Digital & Social Media Strategy at U...
Digital Advertising Lecture for Advanced Digital & Social Media Strategy at U...Digital Advertising Lecture for Advanced Digital & Social Media Strategy at U...
Digital Advertising Lecture for Advanced Digital & Social Media Strategy at U...
 
Invezz.com - Grow your wealth with trading signals
Invezz.com - Grow your wealth with trading signalsInvezz.com - Grow your wealth with trading signals
Invezz.com - Grow your wealth with trading signals
 
BigBuy dropshipping via API with DroFx.pptx
BigBuy dropshipping via API with DroFx.pptxBigBuy dropshipping via API with DroFx.pptx
BigBuy dropshipping via API with DroFx.pptx
 

IBM HR Analytics Employee Attrition & Performance

  • 1. IBM HR Analytics Employee Attrition & Performance Data Science Project by: Angelin Grace Wijaya, Agarwala Pratham, Krishna Shivangi
  • 2. SOURCE - Kaggle Dataset by pavansubhash - Fictional dataset created by IBM data scientists - Contains 35 columns and 1470 records - No missing values https://www.kaggle.com/pavansubhasht/ibm-hr-analytics-attrition-dataset
  • 3. Business Objective of our Project - HR Analysis tool for companies especially those with large workforce. - Help companies to be prepared for future employee-loss - To find possible reasons for employee attrition, in order to prevent valuable employees from leaving. - Predicting employee’s Performance Rating and hence distributing the employees into two classes ( 0 : Low Performance Rating, 1 : High Performance Rating) - Accordingly, predict the employees “Percent Salary Hike” for the next working term.
  • 4. DATA CLEANING Two of the columns in the employee dataset are labelled: “Standard Hours” and “Employee Count”. In “Standard Hours”, we found that all the records have a value of 80. What does this mean? This standard hour is also called a 9/80 work schedule. Here, employees work for 80 hours over a period of 9 days. In “Employee Count”, we found that all the records have a value of 1. What does this mean? Employee count is also known as headcount. It is the number of employees working in the company, where each employee, whether full-time or part-time, is counted as 1. As these columns all have the same values, we decided to drop them while keeping in mind of what each of them represents.
  • 5. DATA CLEANING Two of the columns in the employee dataset are labelled: “Over18” and “NumCompaniesWorked”. In “Over18”, we found that all the records have a value of “Yes”. We found that there are actually 8 employees who are 18 from the “Age” column. This means that the “Over18” column has mismatched values, or that this column actually means employees who are 18 and above. We kept this column and change the records for those who are aged 18 to “No” in their “Over 18” column. In “NumCompaniesWorked”, we found that there are records with zero companies worked. We found that for these employees, they all have 1 year where they worked not for a company (ex. part-time). As 18 year-old employees, who have zero years of work experience, have one as number of companies worked (their current one), we also decided to change the records of number of companies worked as zero to one.
  • 7. JOBS Sales department: - Sales Executive - Sales Representative - Sales Manager Research & Development department: - Research Scientist - Laboratory Technician - Manufacturing Director - Healthcare Representative - Research Director - Research Manager Human Resources department: - Human Resources - Human Resources Manager
  • 8. JOB LEVELS It is defined as the classification of authority and/or salary levels of positions in an organization. Ex. senior executives, executive, middle management, advisors, associate Different job levels have different monthly income.
  • 9. JOB ROLE & AGE - Managers and Research Directors have the oldest median age. - Managers and Research Directors are also earning the highest median monthly income - Sales representatives are the youngest and they are receiving the lowest monthly income
  • 10. JOB ROLE WITH TOTAL WORKING YEARS & YEARS IN CURRENT ROLE - Managers and research directors have the highest total working years and years at current company - Sales representatives have the lowest total working years and years at current company - Years since last promotion is more or less the same for all the job roles other than managers
  • 12. WHAT IS ATTRITION? Attrition is the decrease in the number of employees over time as they leave due to personal reasons or retire. Low attrition rate is desired by many companies because it implies that employees are satisfied with their workplace and the company does not have to train new recruits. Disadvantages of Attrition: 1. Decrease Overall Performance 2. Difficulty in Managing Task (that was left by the employee) 3. Increased Cost (must hire and train new employee) 4. Lack of Knowledgeable/Experienced Employee 5. Losing Out on Employee Development Plan 6. Negative Image
  • 13. DECISION TREE We found that our dataset is imbalanced. The ratio of the number of people with attrition to without attrition is 1233 : 237 The confusion matrix above is before balancing. Those who actually have attrition are being classified as having no attrition. Below is the matrix after balancing. Accuracy Train: 0.86 Accuracy Test: 0.83 Accuracy Train: 0.74 Accuracy Test: 0.77
  • 14. RANDOM FOREST We used GridSearchCV to find the ideal number of trees and depth for our random forest. Result: {'max_depth': 7, 'n_estimators': 100} Below is our confusion matrix for random forest: Accuracy Train: 0.95 Accuracy Test: 0.90 Cross-validation is performed by the GridsearchCV. We pass the whole dataset to it without splitting it and by default it performs 5-fold cross validation Why random forest? A single decision tree like the one in the previous slide brings a much lower accuracy than forest. In random forest, we can plot multiple trees that learn from one another’s mistake (or wrong prediction) to give a better accuracy.
  • 15. ROC CURVES We also found the Best Threshold to be 0.498017 and optimal G-Mean as 0.951
  • 16. VARIABLE IMPORTANCE 1. Overtime 2. Monthly Income 3. Years At Company 7. Total Working Years 8. Stock Option Level 9. Years With Current Manager 4. Job Level 5. Age 6. Years In Current Role Train Test Split has been fixed with random state of 1. Without random state, these 9 variables are still in the top 9 importance but may be in different ranks. (ex. Monthly Income may become more important than Overtime, but is still within the 9 ranks)
  • 17. Attrition and Monthly Income Employees with attrition have lower monthly income Attrition also has a -0.159840 correlation with Monthly Income
  • 18. MONTHLY INCOME- MORE INSIGHTS First of all, Monthly Income is a continuous data, so we performed extensive EDA to study the patterns in Monthly Income . 1. RFECV : Recursive Feature Elimination from sklearn.feature_selection an for CV we will will use KFold validation. 2. Greedy Search : Iteratively removing variables from the dataset until the there are only five variables left excluding the response variable(MonthlyIncome) and criterion is the accuracy score of the model . It removes the variables that least increases the accuracy score .
  • 19. MONTHLY INCOME - RFECV Using RFECV(Recursive Feature Elimination CVross Validation) from sklearn.feature_slection we performed the following the steps: 1. We label-encoded the variables which have some kind of order or have classes 2 or less, and for rest we did One-Hot Encoding. 2. Removing correlated features is done because we don’t want highly correlated features in our dataset as they provide the same information — one is enough. For this step we wrote a simple algorithm. We iterated through the entire correlation matrix and removed variables with correlation with other variables exceeding 0.75. 3. We declared two variables — X and target where first represents all the features, and the second represents the target variable. Then we’ll use an Machine learning Model (We are using LinearRegression). Then we ran RFECV. Here are the arguments: a. Estimator : LinearRegression() b. Cv : KFold(10) c. Scoring : ‘ Accuracy’ 4. Now seeing the graph the optimal of feature is 8 to get Best accuracy for LinearRegression Model. ➔ TotalWorkingYears ➔ Attrition ➔ Department ➔ JobRole ➔ Over18 ➔ BusinessTravel_Non-Travel ➔ EducationField_Human Resources ➔ 'EducationField_Marketing Variables Short-listed after using RFECV:
  • 20. Attrition and Years (1) Employees with attrition have lower years with current manager and total working years Attrition has a -0.156199 correlation with Years With Current Manager and a -0.171063 correlation with Total Working Years
  • 21. Age and Total working years: ● The age has been classified as this: - 20s → younger employees - 40s → middle aged employees - 50s → older employees ● The correlation between total working years and age is 0.68 ● For all three categories, the number of employees who have worked for a shorter period of time are the ones who want to attire ● As the age increases, the total number of working years increases
  • 22. Attrition and Years (2) Employees with attrition have lower years at company and lower years in current role Attrition has a -0.134392 correlation with Years At Company and a -0.160545 correlation with Years In Current Role
  • 23. Among attiring employees, - Sales Manager and Research Director are attiring after longer years at company. - We know that Managers and Research Directors have the oldest median age and the highest median monthly income. Therefore, it is highly possible that they are attiring due to retirement. - Employees that attire with the least number of working years are sales representatives. - Other job roles are attiring after shorter working years, and we predict that it is to look for better job opportunities.
  • 24. Attrition and Job Level Employees with a lower job level are more likely to attire Attrition has a -0.169105 correlation with Job Level
  • 25. Job level and Monthly Income - The correlation between job level and monthly income is 0.95 - Employees with a job level of 5 are either managers and research directors (highest paid) - Employees with a job level of 1 are sales representatives (lowest paid) - As the job level increases, the number of employees that want to attire decrease - Job level of 1 has the highest number of employees that want to attire
  • 26. From this boxplot, we can see how every job level almost has a corresponding range of monthly income. The difference in median monthly income (No Attrition - Attrition ) is greatest for Job Level= 4 followed by Job Level= 1. A possible reason for high attrition rate at Job Level 1 may be that they feel it is unfair to be receiving a lower monthly income than others in the same job level. How about Job level 2 then? The monthly income for both are almost equal, so what could be the reason for them attiring? Also, the difference in monthly income in level 4 is very high, why is the attrition rate lower?
  • 27. Possible reason for high attrition in Job Level 2: - Even though they have nearly the same median monthly income, it turns out that employees here with attrition have a lower median stock option level compared to those not attiring. Possible reason for low attrition in Job Level 4: - Even though there is a huge gap in median monthly income, most of the employees have a high monthly income that may compensate for a lower stock option level. The red, box for those very few employees with attrition have no stocks and a much low monthly income compared to the rest.
  • 28. Attrition and Age There are more younger people with attrition Attrition has a -0.159205 correlation with Age
  • 29. Age was split into 3 categories: ● Younger → aged between 18 and 30, inclusive. There is a 0.259 probability of attiring ● Middle → aged between 31 and 42 inclusive. There is a 0.132 probability of attiring ● Older → aged between 43 and 60 inclusive. There is a 0.116 probability of attiring This grouping aided us in understanding from which age group are most attired employees found. These are the graphs of age against attrition: younger middle older There is a higher probability for the younger group aged 18 to 30 to attire than other age groups. Employees may attire to look for better job opportunities outside the company while they are young.
  • 30. Attrition and Overtime Employees with attrition are more likely those that do not have overtime Attrition has a 0.246118 correlation with Overtime
  • 31. Employees without attrition, whether working overtime or not have a median job satisfaction at 3. However, attiring employees not working overtime are actually less satisfied with their job than those working overtime. median = 3 median = 2 median = 3 Q1= 2 Q1= 1 Q1= 2 Q3= 4
  • 32. In this box plot, we can see how job level 1 employees with attrition and no overtime have a much lower median monthly income for every job involvement value. This could mean that attiring employees feel unfair because they are being paid less than people in the same job level who are also not working overtime and with the same job involvement. This trend is observed for all the other job levels in our Jupyter notebook
  • 33. Attrition and Stock Option Level Employees with attrition are most likely those without any stock option Attrition has a -0.137145 correlation with Stock Option Level
  • 34. For every job role except Human Resources, there seems to be a trend where employees with attrition are those that have a lower stock level compared to those without attrition. Employees might be attiring because they have a lower stock option level than other employees of the same role.
  • 35. PUTTING IT ALL TOGETHER We used Plotly to produce this density heatmap. Most employees with attrition have a low Job Level, low Years at Company, low Stock Option level, and low Monthly Income
  • 37. WHAT IS PERFORMANCE RATING? WHAT ARE WE PLANNING TO DO WITH IT? - Performance Rating is a multi-class variable which partitions the employees into 4 groups. a. Performance Rating of 1 b. Performance Rating of 2 c. Performance Rating of 3 d. Performance Rating of 4 - However we found that the employees only have either a performance rating of 3 or 4 - Thus, we took performance rating as a binary-class variable - Our assumption is supported by the adjacent graph. Which clearly shows, company provides higher increase in salary to employees with Performance Rating of 4. - We will use the predicted values of PerformanceRating to predict the values of PercentSalaryHike of respective employees.
  • 38. DECISION TREE Performance Rating was highly imbalanced so we upsampled the data using “upsample” from sklearn.utils After balancing the data we ran GridSearchCV for tuning max_depth (found it to be 10) and Before Balancing Accuracy: 0.85 After Balancing Train After Balancing Test Accuracy: 0.90 Accuracy: 0.79
  • 39. RANDOM FOREST & VARIABLE IMPORTANCE We used GridSearchCV to find the ideal number of trees and depth for our random forest. RandomForestClassifier(max_depth=9, n_estimators=150) After Balancing Test Accuracy: 0.95
  • 40. PREDICTING PERCENT SALARY HIKE USING PREDICTED VALUES OF PERFORMANCE RATING We know that PercentSalaryHike and PerformanceRating are perfectly correlated in terms of classification, whereas linearly has a correlation of 0.77. So, we can treat PercentSalaryHike in two ways : ● As a continuous variable and perform a regression model. ○ RandomForestRegressor ● As a multi-class variable and perform a multi-classification model. ○ RandomForestClassifier
  • 41. RandomForestClassifier RandomForestRegressor RandomForestClassifier Accuracy Test: 0.841 Classification Accuracy Test : 0.55 So comparing the accuracy of the models we decided to go with the regression model to predict PercentSalaryHike
  • 42. “Research indicates that workers have three prime needs: Interesting work, recognition for doing a good job, and being let in on things that are going on in the company.” - Zig Ziglar
  • 43. CONCLUSION To have a low attrition rate, it is important for companies to: 1. Fairly pay employees in same job level, same job involvement, same job role with almost equal monthly income and stocks. 2. Offer stock options to more employees especially new recruits as an incentive. 3. Increase allowance for employees to work overtime, and make sure that those working overtime are being paid more than those who are not. 4. Also from our quote, companies should remember to appreciate their employees and work on the harmonic balance of their company.