---
title: "Meridian Group"
subtitle: "People Analytics: Employee Attrition Intelligence Report"
author: "Freda Erinmwingbovo, Data Analyst"
date: today
format:
html:
theme: [flatly, meridian.scss]
toc: true
toc-depth: 3
toc-title: "Report Contents"
toc-location: left
number-sections: false
code-fold: true
code-tools: true
code-summary: "Show Code"
page-layout: full
smooth-scroll: true
self-contained: true
execute:
warning: false
message: false
cache: true
jupyter: python3
jupyter-options:
python: "C:/Users/engrf/AppData/Local/Programs/Python/Python313/python.exe"
---
# Meridian Group: People Analytics Report
**Prepared by:** Freda Erinmwingbovo, Data Analyst
**Prepared for:** Dr. Amara Osei, Chief Executive Officer, Meridian Group
**Date:** February 2026
**Dataset:** 1,470 employees · 35 attributes · IBM HR Analytics
---
## Executive Summary
Meridian Group employed **1,470 people** at the time of this analysis.
Of those, **237 have left the organisation**, an attrition rate of
**16.1%**, above the industry benchmark of 10 to 12% for professional
services organisations of comparable size.
This report delivers a seven-theme people analytics investigation
identifying who is leaving, why, and what the organisation must do
to reverse the trend.
```{=html}
<div class="kpi-grid">
<div class="kpi-card">
<div class="kpi-value">1,470</div>
<div class="kpi-label">Total Employees</div>
</div>
<div class="kpi-card">
<div class="kpi-value alert">237</div>
<div class="kpi-label">Employees Left</div>
</div>
<div class="kpi-card">
<div class="kpi-value alert">16.1%</div>
<div class="kpi-label">Attrition Rate</div>
</div>
<div class="kpi-card">
<div class="kpi-value">$6,503</div>
<div class="kpi-label">Avg Monthly Income</div>
</div>
<div class="kpi-card">
<div class="kpi-value">7.0 yrs</div>
<div class="kpi-label">Avg Tenure</div>
</div>
</div>
```
| # | Business Question | Priority |
|---|------------------|----------|
| 1 | Overall attrition profile | Baseline |
| 2 | Department and role attrition | High |
| 3 | Compensation and attrition | High |
| 4 | Satisfaction scores and attrition | High |
| 5 | Overtime impact on attrition | Critical |
| 6 | Tenure and career progression | High |
| 7 | Demographic profile of leavers | Medium |
---
## Data and Methodology
### Data Loading and Preparation
```{python}
import pandas as pd
import numpy as np
import matplotlib
import matplotlib.pyplot as plt
import matplotlib.ticker as mticker
import warnings
warnings.filterwarnings('ignore')
df = pd.read_csv('hr-employee-attrition.csv')
# Derived columns
df['AttritionBinary'] = (df['Attrition'] == 'Yes').astype(int)
# Satisfaction label maps
sat_map = {1: 'Low', 2: 'Medium', 3: 'High', 4: 'Very High'}
df['JobSatisfactionLabel'] = df['JobSatisfaction'].map(sat_map)
df['EnvSatisfactionLabel'] = df['EnvironmentSatisfaction'].map(sat_map)
df['RelSatisfactionLabel'] = df['RelationshipSatisfaction'].map(sat_map)
df['WLBLabel'] = df['WorkLifeBalance'].map(sat_map)
# Education map
edu_map = {1:'Below College', 2:'College', 3:'Bachelor',
4:'Master', 5:'Doctor'}
df['EducationLabel'] = df['Education'].map(edu_map)
# Age bands
def age_band(age):
if age < 25: return 'Under 25'
elif age < 35: return '25 to 34'
elif age < 45: return '35 to 44'
elif age < 55: return '45 to 54'
else: return '55 and over'
df['AgeBand'] = df['Age'].apply(age_band)
# Income bands
def income_band(inc):
if inc < 3000: return 'Under $3K'
elif inc < 6000: return '$3K to $6K'
elif inc < 10000: return '$6K to $10K'
elif inc < 15000: return '$10K to $15K'
else: return 'Over $15K'
df['IncomeBand'] = df['MonthlyIncome'].apply(income_band)
# Drop constant columns
df = df.drop(columns=['EmployeeCount', 'Over18', 'StandardHours'])
print(f"Dataset ready: {df.shape[0]:,} employees · {df.shape[1]} attributes")
print(f"\nAttrition breakdown:")
print(df['Attrition'].value_counts())
print(f"\nAttrition rate: {df['AttritionBinary'].mean()*100:.1f}%")
```
### Portfolio Overview
```{python}
print("MERIDIAN GROUP WORKFORCE OVERVIEW")
print("=" * 45)
print(f"Total employees: {len(df):,}")
print(f"Employees who left: {df['AttritionBinary'].sum():,}")
print(f"Current employees: {(df['AttritionBinary']==0).sum():,}")
print(f"Attrition rate: {df['AttritionBinary'].mean()*100:.1f}%")
print(f"Departments: {df['Department'].nunique()}")
print(f"Job roles: {df['JobRole'].nunique()}")
print(f"Avg age: {df['Age'].mean():.1f} years")
print(f"Avg monthly income: ${df['MonthlyIncome'].mean():,.0f}")
print(f"Avg tenure: {df['YearsAtCompany'].mean():.1f} years")
print(f"Overtime workers: "
f"{(df['OverTime']=='Yes').sum():,} "
f"({(df['OverTime']=='Yes').mean()*100:.1f}%)")
```
---
# Analysis
::: {.panel-tabset}
## Attrition Overview
### Business Question
What is the overall attrition picture at Meridian Group?
How does attrition vary across departments and what does
the leaver profile look like at the highest level?
### Overall Attrition
```{python}
dept_attr = df.groupby('Department').agg(
Total = ('AttritionBinary', 'count'),
Left = ('AttritionBinary', 'sum'),
Stayed = ('AttritionBinary', lambda x: (x==0).sum())
).reset_index()
dept_attr['AttritionRate'] = (
dept_attr['Left'] / dept_attr['Total'] * 100
).round(1)
display(dept_attr)
```
### Attrition Overview Visualization
```{python}
navy = '#1a2744'
teal = '#2a7a6f'
amber = '#d4722a'
red = '#b83232'
muted = '#7a7468'
bg = '#f5f4f0'
white = '#ffffff'
border = '#ddd9d0'
fig, axes = plt.subplots(1, 3, figsize=(16, 6))
fig.patch.set_facecolor(bg)
fig.suptitle('Figure: Meridian Group Attrition Overview',
color=navy, fontsize=13,
fontfamily='serif', fontweight='bold', y=1.02)
# Plot 1: Overall attrition donut
ax1 = axes[0]
ax1.set_facecolor(bg)
sizes = [1233, 237]
colors = [teal, red]
labels = ['Active\n1,233', 'Left\n237']
wedges, texts = ax1.pie(sizes, colors=colors,
startangle=90,
wedgeprops={'width': 0.55,
'edgecolor': white,
'linewidth': 2})
ax1.text(0, 0, '16.1%\nAttrition',
ha='center', va='center',
fontsize=12, fontfamily='serif',
color=navy, fontweight='bold')
ax1.set_title('Overall Attrition Rate',
color=navy, fontweight='bold',
pad=12, fontfamily='serif')
ax1.legend(labels, loc='lower center',
fontsize=9, frameon=False,
labelcolor=navy)
# Plot 2: Attrition rate by department
ax2 = axes[1]
ax2.set_facecolor(white)
d_colors = [red if r == dept_attr['AttritionRate'].max()
else teal for r in dept_attr['AttritionRate']]
bars = ax2.bar(dept_attr['Department'],
dept_attr['AttritionRate'],
color=d_colors, edgecolor='none', width=0.5)
ax2.set_ylabel('Attrition Rate (%)', color=muted)
ax2.set_title('Attrition Rate by Department',
color=navy, fontweight='bold',
pad=12, fontfamily='serif')
ax2.tick_params(colors=muted)
for spine in ax2.spines.values():
spine.set_color(border)
ax2.grid(color=border, linestyle='--',
linewidth=0.5, axis='y')
for bar, val in zip(bars, dept_attr['AttritionRate']):
ax2.text(bar.get_x() + bar.get_width()/2,
bar.get_height() + 0.3,
f'{val}%',
ha='center', color=navy,
fontsize=10, fontweight='bold')
ax2.set_xticklabels(dept_attr['Department'],
rotation=15, ha='right',
color=muted, fontsize=8)
# Plot 3: Headcount by department
ax3 = axes[2]
ax3.set_facecolor(white)
x = range(len(dept_attr))
bars_stayed = ax3.bar(x, dept_attr['Stayed'],
color=teal, edgecolor='none',
width=0.5, label='Active')
bars_left = ax3.bar(x, dept_attr['Left'],
bottom=dept_attr['Stayed'],
color=red, edgecolor='none',
width=0.5, label='Left',
alpha=0.85)
ax3.set_xticks(x)
ax3.set_xticklabels(dept_attr['Department'],
rotation=15, ha='right',
color=muted, fontsize=8)
ax3.set_ylabel('Headcount', color=muted)
ax3.set_title('Headcount by Department',
color=navy, fontweight='bold',
pad=12, fontfamily='serif')
ax3.tick_params(colors=muted)
for spine in ax3.spines.values():
spine.set_color(border)
ax3.legend(facecolor=bg, edgecolor=border,
fontsize=9, labelcolor=navy)
plt.tight_layout()
plt.show()
```
### Attrition Overview Findings
Research and Development is the largest department and carries
a significant share of total attrition. Sales shows the highest
attrition rate, representing a priority intervention target.
**Recommendations:**
- Launch an immediate retention investigation in the
Sales department. The elevated attrition rate signals
systemic issues beyond individual decisions
- Benchmark Meridian Group's 16.1% rate against the
professional services sector average of 10 to 12%
and set a formal 12-month reduction target
---
## Department and Role Analysis
### Business Question
Which specific job roles carry the highest attrition rates?
Does job level influence the likelihood of leaving?
### Attrition by Job Role
```{python}
role_attr = df.groupby('JobRole').agg(
Total = ('AttritionBinary', 'count'),
Left = ('AttritionBinary', 'sum'),
AvgIncome = ('MonthlyIncome', 'mean'),
AvgTenure = ('YearsAtCompany', 'mean')
).reset_index()
role_attr['AttritionRate'] = (
role_attr['Left'] / role_attr['Total'] * 100
).round(1)
role_attr = role_attr.sort_values(
'AttritionRate', ascending=False
)
display(role_attr)
```
### Attrition by Job Level
```{python}
level_attr = df.groupby('JobLevel').agg(
Total = ('AttritionBinary', 'count'),
Left = ('AttritionBinary', 'sum'),
AvgIncome = ('MonthlyIncome', 'mean')
).reset_index()
level_attr['AttritionRate'] = (
level_attr['Left'] / level_attr['Total'] * 100
).round(1)
level_attr['JobLevelLabel'] = level_attr['JobLevel'].map({
1:'Level 1\n(Junior)', 2:'Level 2\n(Mid)',
3:'Level 3\n(Senior)', 4:'Level 4\n(Lead)',
5:'Level 5\n(Director)'
})
display(level_attr)
```
### Role and Level Visualization
```{python}
fig, axes = plt.subplots(1, 2, figsize=(16, 7))
fig.patch.set_facecolor(bg)
fig.suptitle('Figure II: Attrition by Job Role and Level',
color=navy, fontsize=13,
fontfamily='serif', fontweight='bold', y=1.02)
# Plot 1: Attrition rate by role
ax1 = axes[0]
ax1.set_facecolor(white)
r_colors = [red if r == role_attr['AttritionRate'].max()
else teal for r in role_attr['AttritionRate']]
bars = ax1.barh(range(len(role_attr)),
role_attr['AttritionRate'],
color=r_colors, edgecolor='none', height=0.65)
ax1.set_yticks(range(len(role_attr)))
ax1.set_yticklabels(role_attr['JobRole'],
color=muted, fontsize=9)
ax1.set_xlabel('Attrition Rate (%)', color=muted)
ax1.set_title('Attrition Rate by Job Role',
color=navy, fontweight='bold',
pad=12, fontfamily='serif')
ax1.tick_params(colors=muted)
for spine in ax1.spines.values():
spine.set_color(border)
ax1.grid(color=border, linestyle='--',
linewidth=0.5, axis='x')
ax1.invert_yaxis()
for bar, val in zip(bars, role_attr['AttritionRate']):
ax1.text(bar.get_width() + 0.2,
bar.get_y() + bar.get_height()/2,
f'{val}%', va='center',
color=navy, fontsize=8,
fontweight='bold')
# Plot 2: Attrition rate by job level
ax2 = axes[1]
ax2.set_facecolor(white)
l_colors = [red if r == level_attr['AttritionRate'].max()
else teal for r in level_attr['AttritionRate']]
bars2 = ax2.bar(level_attr['JobLevelLabel'],
level_attr['AttritionRate'],
color=l_colors, edgecolor='none', width=0.5)
ax2.set_ylabel('Attrition Rate (%)', color=muted)
ax2.set_title('Attrition Rate by Job Level',
color=navy, fontweight='bold',
pad=12, fontfamily='serif')
ax2.tick_params(colors=muted)
for spine in ax2.spines.values():
spine.set_color(border)
ax2.grid(color=border, linestyle='--',
linewidth=0.5, axis='y')
for bar, val in zip(bars2, level_attr['AttritionRate']):
ax2.text(bar.get_x() + bar.get_width()/2,
bar.get_height() + 0.3,
f'{val}%', ha='center',
color=navy, fontsize=10,
fontweight='bold')
plt.tight_layout()
plt.show()
```
### Role and Level Findings and Recommendations
Junior level employees (Level 1) carry a significantly higher
attrition rate than any other level, a pattern consistent with
inadequate onboarding, limited career visibility, and
compensation gaps at entry level.
**Recommendations:**
- Implement structured 90-day and 180-day onboarding
reviews for all Level 1 employees
- Create visible career pathways from Level 1 to Level 2
with clear criteria and timelines
- Review compensation benchmarks for the highest-attrition
job roles against external market data
---
## Compensation Analysis
### Business Question
Is there a measurable relationship between monthly income,
salary hike percentage, stock options, and the likelihood
of leaving Meridian Group?
### Income by Attrition Status
```{python}
income_attr = df.groupby(['IncomeBand', 'Attrition']).size()\
.reset_index(name='Count')
income_pivot = income_attr.pivot(
index='IncomeBand', columns='Attrition', values='Count'
).fillna(0).reset_index()
income_pivot['Total'] = income_pivot['No'] + income_pivot['Yes']
income_pivot['AttritionRate'] = (
income_pivot['Yes'] / income_pivot['Total'] * 100
).round(1)
display(income_pivot)
```
### Salary Hike and Stock Options vs Attrition
```{python}
hike_attr = df.groupby('Attrition').agg(
AvgSalaryHike = ('PercentSalaryHike', 'mean'),
AvgStockOption = ('StockOptionLevel', 'mean'),
AvgMonthlyIncome = ('MonthlyIncome', 'mean')
).round(2).reset_index()
display(hike_attr)
```
### Compensation Visualization
```{python}
fig, axes = plt.subplots(1, 2, figsize=(16, 6))
fig.patch.set_facecolor(bg)
fig.suptitle('Figure III: Compensation and Attrition',
color=navy, fontsize=13,
fontfamily='serif', fontweight='bold', y=1.02)
# Plot 1: Income band attrition rate
income_order = ['Under $3K', '$3K to $6K',
'$6K to $10K', '$10K to $15K', 'Over $15K']
ip = income_pivot.set_index('IncomeBand').reindex(
income_order
).reset_index()
ax1 = axes[0]
ax1.set_facecolor(white)
i_colors = [red if r == ip['AttritionRate'].max()
else teal for r in ip['AttritionRate']]
bars = ax1.bar(range(len(income_order)),
ip['AttritionRate'],
color=i_colors, edgecolor='none', width=0.6)
ax1.set_xticks(range(len(income_order)))
ax1.set_xticklabels(income_order, rotation=20,
ha='right', color=muted, fontsize=8)
ax1.set_ylabel('Attrition Rate (%)', color=muted)
ax1.set_title('Attrition Rate by Income Band',
color=navy, fontweight='bold',
pad=12, fontfamily='serif')
ax1.tick_params(colors=muted)
for spine in ax1.spines.values():
spine.set_color(border)
ax1.grid(color=border, linestyle='--',
linewidth=0.5, axis='y')
for bar, val in zip(bars, ip['AttritionRate']):
ax1.text(bar.get_x() + bar.get_width()/2,
bar.get_height() + 0.3,
f'{val}%', ha='center',
color=navy, fontsize=9,
fontweight='bold')
# Plot 2: Avg income: leavers vs stayers
ax2 = axes[1]
ax2.set_facecolor(white)
groups = hike_attr['Attrition'].tolist()
incomes = hike_attr['AvgMonthlyIncome'].tolist()
c2 = [red if g == 'Yes' else teal for g in groups]
bars2 = ax2.bar(['Left', 'Stayed'], incomes,
color=[red, teal],
edgecolor='none', width=0.5)
ax2.set_ylabel('Average Monthly Income ($)', color=muted)
ax2.set_title('Average Income: Leavers vs Stayers',
color=navy, fontweight='bold',
pad=12, fontfamily='serif')
ax2.tick_params(colors=muted)
for spine in ax2.spines.values():
spine.set_color(border)
ax2.grid(color=border, linestyle='--',
linewidth=0.5, axis='y')
for bar, val in zip(bars2, incomes):
ax2.text(bar.get_x() + bar.get_width()/2,
bar.get_height() + 50,
f'${val:,.0f}',
ha='center', color=navy,
fontsize=11, fontweight='bold')
plt.tight_layout()
plt.show()
```
### Compensation Findings and Recommendations
Employees who leave earn significantly less on average than
those who stay. The Under $3K income band carries the highest
attrition rate, confirming that compensation is a primary
attrition driver at the lower end of the pay scale.
**Recommendations:**
- Conduct an immediate compensation review for all
employees in the Under $3K and $3K to $6K income bands
- Introduce a retention bonus scheme for high-performers
in the Under $3K band who have been in role for
more than 12 months
- Review stock option allocation policy, employees
with zero stock options show higher attrition rates
---
## Satisfaction and Wellbeing
### Business Question
Do the four satisfaction dimensions, job satisfaction,
environment satisfaction, work-life balance, and relationship
satisfaction, predict attrition at Meridian Group?
### Satisfaction Scores by Attrition Status
```{python}
satisfaction_summary = df.groupby('Attrition').agg(
AvgJobSatisfaction = ('JobSatisfaction', 'mean'),
AvgEnvSatisfaction = ('EnvironmentSatisfaction', 'mean'),
AvgRelSatisfaction = ('RelationshipSatisfaction', 'mean'),
AvgWorkLifeBalance = ('WorkLifeBalance', 'mean')
).round(3).reset_index()
display(satisfaction_summary)
```
### Low Satisfaction Attrition Rates
```{python}
low_sat = df[df['JobSatisfaction'] == 1]
low_env = df[df['EnvironmentSatisfaction'] == 1]
low_wlb = df[df['WorkLifeBalance'] == 1]
print("Attrition rate among LOW satisfaction employees:")
print(f" Low Job Satisfaction (score=1): "
f"{low_sat['AttritionBinary'].mean()*100:.1f}% "
f"({low_sat['AttritionBinary'].sum()} of {len(low_sat)})")
print(f" Low Environment Satisfaction (score=1): "
f"{low_env['AttritionBinary'].mean()*100:.1f}% "
f"({low_env['AttritionBinary'].sum()} of {len(low_env)})")
print(f" Poor Work-Life Balance (score=1): "
f"{low_wlb['AttritionBinary'].mean()*100:.1f}% "
f"({low_wlb['AttritionBinary'].sum()} of {len(low_wlb)})")
```
### Satisfaction Visualization
```{python}
fig, axes = plt.subplots(2, 2, figsize=(16, 10))
fig.patch.set_facecolor(bg)
fig.suptitle('Figure IV: Satisfaction Scores and Attrition',
color=navy, fontsize=13,
fontfamily='serif', fontweight='bold', y=1.02)
sat_cols = [
('JobSatisfaction', 'Job Satisfaction'),
('EnvironmentSatisfaction', 'Environment Satisfaction'),
('WorkLifeBalance', 'Work-Life Balance'),
('RelationshipSatisfaction', 'Relationship Satisfaction')
]
for idx, (col, label) in enumerate(sat_cols):
ax = axes[idx // 2][idx % 2]
ax.set_facecolor(white)
sat_data = df.groupby([col, 'Attrition']).size()\
.reset_index(name='Count')
sat_pivot = sat_data.pivot(
index=col, columns='Attrition', values='Count'
).fillna(0)
sat_pivot['AttritionRate'] = (
sat_pivot['Yes'] /
(sat_pivot['Yes'] + sat_pivot['No']) * 100
).round(1)
s_colors = [red if r == sat_pivot['AttritionRate'].max()
else teal
for r in sat_pivot['AttritionRate']]
bars = ax.bar(sat_pivot.index,
sat_pivot['AttritionRate'],
color=s_colors, edgecolor='none',
width=0.55)
ax.set_xlabel('Score (1=Low, 4=Very High)',
color=muted, fontsize=8)
ax.set_ylabel('Attrition Rate (%)', color=muted)
ax.set_title(label, color=navy,
fontweight='bold', pad=10,
fontfamily='serif')
ax.tick_params(colors=muted)
for spine in ax.spines.values():
spine.set_color(border)
ax.grid(color=border, linestyle='--',
linewidth=0.5, axis='y')
for bar, val in zip(bars,
sat_pivot['AttritionRate']):
ax.text(bar.get_x() + bar.get_width()/2,
bar.get_height() + 0.3,
f'{val}%', ha='center',
color=navy, fontsize=9,
fontweight='bold')
plt.tight_layout()
plt.show()
```
### Satisfaction Findings and Recommendations
Employees with the lowest job satisfaction and environment
satisfaction scores show materially higher attrition rates.
Poor work-life balance is the single strongest satisfaction
predictor of attrition in this dataset.
**Recommendations:**
- Deploy a quarterly pulse survey targeting all four
satisfaction dimensions, score deterioration is
an early warning signal that must trigger manager action
- Introduce a mandatory manager conversation for any
employee who scores 1 on job satisfaction or
work-life balance in consecutive surveys
- Review workload distribution across teams
with the highest concentration of overtime workers
---
## Overtime Impact
### Business Question
Is overtime a significant driver of attrition at
Meridian Group? How does the attrition rate compare
between employees who work overtime and those who do not?
### Overtime and Attrition
```{python}
ot_attr = df.groupby('OverTime').agg(
Total = ('AttritionBinary', 'count'),
Left = ('AttritionBinary', 'sum'),
AvgIncome = ('MonthlyIncome', 'mean'),
AvgSatisfaction = ('JobSatisfaction', 'mean'),
AvgWLB = ('WorkLifeBalance', 'mean')
).reset_index()
ot_attr['AttritionRate'] = (
ot_attr['Left'] / ot_attr['Total'] * 100
).round(1)
display(ot_attr)
```
### Overtime by Department
```{python}
ot_dept = df.groupby(['Department', 'OverTime']).agg(
Total = ('AttritionBinary', 'count'),
Left = ('AttritionBinary', 'sum')
).reset_index()
ot_dept['AttritionRate'] = (
ot_dept['Left'] / ot_dept['Total'] * 100
).round(1)
display(ot_dept)
```
### Overtime Visualization
```{python}
fig, axes = plt.subplots(1, 2, figsize=(16, 6))
fig.patch.set_facecolor(bg)
fig.suptitle('Figure V: Overtime and Attrition',
color=navy, fontsize=13,
fontfamily='serif', fontweight='bold', y=1.02)
# Plot 1: Attrition rate overtime vs not
ax1 = axes[0]
ax1.set_facecolor(white)
ot_labels = ot_attr['OverTime'].tolist()
ot_rates = ot_attr['AttritionRate'].tolist()
ot_colors = [red if l == 'Yes' else teal for l in ot_labels]
bars = ax1.bar(['No Overtime', 'Works Overtime'],
ot_rates, color=ot_colors,
edgecolor='none', width=0.5)
ax1.set_ylabel('Attrition Rate (%)', color=muted)
ax1.set_title('Attrition Rate: Overtime vs No Overtime',
color=navy, fontweight='bold',
pad=12, fontfamily='serif')
ax1.tick_params(colors=muted)
for spine in ax1.spines.values():
spine.set_color(border)
ax1.grid(color=border, linestyle='--',
linewidth=0.5, axis='y')
for bar, val in zip(bars, ot_rates):
ax1.text(bar.get_x() + bar.get_width()/2,
bar.get_height() + 0.3,
f'{val}%', ha='center',
color=navy, fontsize=14,
fontweight='bold')
# Plot 2: Overtime headcount by department
ax2 = axes[1]
ax2.set_facecolor(white)
ot_yes = ot_dept[ot_dept['OverTime'] == 'Yes']
ot_no = ot_dept[ot_dept['OverTime'] == 'No']
depts = ot_dept['Department'].unique()
x = range(len(depts))
bars_no = ax2.bar([i - 0.2 for i in x],
ot_no['Total'],
width=0.35, color=teal,
edgecolor='none', label='No Overtime')
bars_yes = ax2.bar([i + 0.2 for i in x],
ot_yes['Total'],
width=0.35, color=amber,
edgecolor='none', label='Works Overtime')
ax2.set_xticks(x)
ax2.set_xticklabels(depts, rotation=15,
ha='right', color=muted, fontsize=8)
ax2.set_ylabel('Headcount', color=muted)
ax2.set_title('Overtime Distribution by Department',
color=navy, fontweight='bold',
pad=12, fontfamily='serif')
ax2.tick_params(colors=muted)
for spine in ax2.spines.values():
spine.set_color(border)
ax2.legend(facecolor=bg, edgecolor=border,
fontsize=9, labelcolor=navy)
plt.tight_layout()
plt.show()
```
### Overtime Findings and Recommendations
Overtime is the single strongest predictor of attrition
in this dataset. Employees who work overtime leave at
a rate more than three times higher than those who do not.
**Recommendations:**
- Implement a mandatory overtime monitoring policy.
No employee should work overtime for more than
eight consecutive weeks without a formal review
- Audit all roles with high overtime incidence for
headcount adequacy, overtime is frequently a
symptom of understaffing, not productivity
- Include overtime hours in the monthly HR dashboard
reviewed by department heads
---
## Tenure and Career Progression
### Business Question
Does time in role, time since last promotion, or total
tenure predict attrition? Are employees leaving because
they feel their careers are stagnating?
### Tenure Profile by Attrition
```{python}
tenure_attr = df.groupby('Attrition').agg(
AvgYearsAtCompany = ('YearsAtCompany', 'mean'),
AvgYearsInRole = ('YearsInCurrentRole', 'mean'),
AvgYearsSincePromotion = ('YearsSinceLastPromotion', 'mean'),
AvgYearsWithManager = ('YearsWithCurrManager', 'mean'),
AvgTotalWorkingYears = ('TotalWorkingYears', 'mean'),
AvgNumCompanies = ('NumCompaniesWorked', 'mean')
).round(2).reset_index()
display(tenure_attr)
```
### Early Attrition: First Two Years
```{python}
early = df[df['YearsAtCompany'] <= 2]
print(f"Employees with 2 years or less tenure: {len(early):,}")
print(f"Attrition rate in this group: "
f"{early['AttritionBinary'].mean()*100:.1f}%")
print(f"vs overall attrition rate: "
f"{df['AttritionBinary'].mean()*100:.1f}%")
```
### Tenure Visualization
```{python}
fig, axes = plt.subplots(1, 2, figsize=(16, 6))
fig.patch.set_facecolor(bg)
fig.suptitle('Figure VI: Tenure and Career Progression',
color=navy, fontsize=13,
fontfamily='serif', fontweight='bold', y=1.02)
# Plot 1: Avg tenure metrics: leavers vs stayers
metrics = ['AvgYearsAtCompany', 'AvgYearsInRole',
'AvgYearsSincePromotion', 'AvgYearsWithManager']
labels = ['Years at\nCompany', 'Years in\nRole',
'Since Last\nPromotion', 'With Current\nManager']
left_vals = tenure_attr[tenure_attr['Attrition']=='Yes'][metrics].values[0]
stayed_vals = tenure_attr[tenure_attr['Attrition']=='No'][metrics].values[0]
x = range(len(metrics))
ax1 = axes[0]
ax1.set_facecolor(white)
bars_left = ax1.bar([i - 0.2 for i in x], left_vals,
width=0.35, color=red,
edgecolor='none', label='Left')
bars_stayed = ax1.bar([i + 0.2 for i in x], stayed_vals,
width=0.35, color=teal,
edgecolor='none', label='Stayed')
ax1.set_xticks(x)
ax1.set_xticklabels(labels, color=muted, fontsize=8)
ax1.set_ylabel('Average Years', color=muted)
ax1.set_title('Tenure Metrics: Leavers vs Stayers',
color=navy, fontweight='bold',
pad=12, fontfamily='serif')
ax1.tick_params(colors=muted)
for spine in ax1.spines.values():
spine.set_color(border)
ax1.grid(color=border, linestyle='--',
linewidth=0.5, axis='y')
ax1.legend(facecolor=bg, edgecolor=border,
fontsize=9, labelcolor=navy)
# Plot 2: Attrition by years at company band
tenure_band = df.copy()
tenure_band['TenureBand'] = pd.cut(
tenure_band['YearsAtCompany'],
bins=[0, 1, 2, 5, 10, 40],
labels=['0-1 yr', '1-2 yrs', '2-5 yrs',
'5-10 yrs', '10+ yrs']
)
tb_attr = tenure_band.groupby('TenureBand').agg(
Total = ('AttritionBinary', 'count'),
Left = ('AttritionBinary', 'sum')
).reset_index()
tb_attr['AttritionRate'] = (
tb_attr['Left'] / tb_attr['Total'] * 100
).round(1)
ax2 = axes[1]
ax2.set_facecolor(white)
tb_colors = [red if r == tb_attr['AttritionRate'].max()
else teal for r in tb_attr['AttritionRate']]
bars2 = ax2.bar(tb_attr['TenureBand'].astype(str),
tb_attr['AttritionRate'],
color=tb_colors, edgecolor='none',
width=0.6)
ax2.set_ylabel('Attrition Rate (%)', color=muted)
ax2.set_title('Attrition Rate by Tenure Band',
color=navy, fontweight='bold',
pad=12, fontfamily='serif')
ax2.tick_params(colors=muted)
for spine in ax2.spines.values():
spine.set_color(border)
ax2.grid(color=border, linestyle='--',
linewidth=0.5, axis='y')
for bar, val in zip(bars2, tb_attr['AttritionRate']):
ax2.text(bar.get_x() + bar.get_width()/2,
bar.get_height() + 0.3,
f'{val}%', ha='center',
color=navy, fontsize=9,
fontweight='bold')
plt.tight_layout()
plt.show()
```
### Tenure Findings and Recommendations
Employees who leave have significantly shorter average tenure
and longer time since last promotion than those who stay.
The first two years represent the highest-risk attrition window.
**Recommendations:**
- Implement a formal promotion review cycle that
guarantees every employee is reviewed for advancement
at 18-month intervals
- Assign every new employee a senior mentor for
their first 24 months, early tenure attrition
is largely preventable
- Flag employees approaching 24 months with no
promotion or role change for a mandatory
career development conversation
---
## Demographic Profile
### Business Question
What is the demographic profile of employees who leave
Meridian Group? Does age, gender, marital status, or
business travel frequency influence attrition?
### Attrition by Gender and Age
```{python}
gender_attr = df.groupby('Gender').agg(
Total = ('AttritionBinary', 'count'),
Left = ('AttritionBinary', 'sum')
).reset_index()
gender_attr['AttritionRate'] = (
gender_attr['Left'] / gender_attr['Total'] * 100
).round(1)
display(gender_attr)
```
```{python}
age_order = ['Under 25', '25 to 34', '35 to 44',
'45 to 54', '55 and over']
age_attr = df.groupby('AgeBand').agg(
Total = ('AttritionBinary', 'count'),
Left = ('AttritionBinary', 'sum')
).reset_index()
age_attr['AttritionRate'] = (
age_attr['Left'] / age_attr['Total'] * 100
).round(1)
age_attr = age_attr.set_index('AgeBand').reindex(
age_order
).reset_index()
display(age_attr)
```
### Attrition by Marital Status and Travel
```{python}
marital_attr = df.groupby('MaritalStatus').agg(
Total = ('AttritionBinary', 'count'),
Left = ('AttritionBinary', 'sum')
).reset_index()
marital_attr['AttritionRate'] = (
marital_attr['Left'] / marital_attr['Total'] * 100
).round(1)
display(marital_attr)
```
```{python}
travel_attr = df.groupby('BusinessTravel').agg(
Total = ('AttritionBinary', 'count'),
Left = ('AttritionBinary', 'sum')
).reset_index()
travel_attr['AttritionRate'] = (
travel_attr['Left'] / travel_attr['Total'] * 100
).round(1)
display(travel_attr)
```
### Demographic Visualization
```{python}
fig, axes = plt.subplots(2, 2, figsize=(16, 10))
fig.patch.set_facecolor(bg)
fig.suptitle('Figure VII: Demographic Attrition Profile',
color=navy, fontsize=13,
fontfamily='serif', fontweight='bold', y=1.02)
# Plot 1: Gender
ax1 = axes[0][0]
ax1.set_facecolor(white)
g_colors = [teal, amber]
bars = ax1.bar(gender_attr['Gender'],
gender_attr['AttritionRate'],
color=g_colors, edgecolor='none',
width=0.45)
ax1.set_ylabel('Attrition Rate (%)', color=muted)
ax1.set_title('Attrition by Gender', color=navy,
fontweight='bold', pad=10,
fontfamily='serif')
ax1.tick_params(colors=muted)
for spine in ax1.spines.values():
spine.set_color(border)
ax1.grid(color=border, linestyle='--',
linewidth=0.5, axis='y')
for bar, val in zip(bars, gender_attr['AttritionRate']):
ax1.text(bar.get_x() + bar.get_width()/2,
bar.get_height() + 0.2,
f'{val}%', ha='center',
color=navy, fontsize=11,
fontweight='bold')
# Plot 2: Age band
ax2 = axes[0][1]
ax2.set_facecolor(white)
a_colors = [red if r == age_attr['AttritionRate'].max()
else teal for r in age_attr['AttritionRate']]
bars2 = ax2.bar(range(len(age_order)),
age_attr['AttritionRate'],
color=a_colors, edgecolor='none',
width=0.6)
ax2.set_xticks(range(len(age_order)))
ax2.set_xticklabels(age_order, rotation=20,
ha='right', color=muted,
fontsize=8)
ax2.set_ylabel('Attrition Rate (%)', color=muted)
ax2.set_title('Attrition by Age Group', color=navy,
fontweight='bold', pad=10,
fontfamily='serif')
ax2.tick_params(colors=muted)
for spine in ax2.spines.values():
spine.set_color(border)
ax2.grid(color=border, linestyle='--',
linewidth=0.5, axis='y')
for bar, val in zip(bars2, age_attr['AttritionRate']):
ax2.text(bar.get_x() + bar.get_width()/2,
bar.get_height() + 0.2,
f'{val}%', ha='center',
color=navy, fontsize=9,
fontweight='bold')
# Plot 3: Marital status
ax3 = axes[1][0]
ax3.set_facecolor(white)
m_colors = [red if r == marital_attr['AttritionRate'].max()
else teal for r in marital_attr['AttritionRate']]
bars3 = ax3.bar(marital_attr['MaritalStatus'],
marital_attr['AttritionRate'],
color=m_colors, edgecolor='none',
width=0.5)
ax3.set_ylabel('Attrition Rate (%)', color=muted)
ax3.set_title('Attrition by Marital Status', color=navy,
fontweight='bold', pad=10,
fontfamily='serif')
ax3.tick_params(colors=muted)
for spine in ax3.spines.values():
spine.set_color(border)
ax3.grid(color=border, linestyle='--',
linewidth=0.5, axis='y')
for bar, val in zip(bars3, marital_attr['AttritionRate']):
ax3.text(bar.get_x() + bar.get_width()/2,
bar.get_height() + 0.2,
f'{val}%', ha='center',
color=navy, fontsize=10,
fontweight='bold')
# Plot 4: Business travel
ax4 = axes[1][1]
ax4.set_facecolor(white)
t_colors = [red if r == travel_attr['AttritionRate'].max()
else teal for r in travel_attr['AttritionRate']]
bars4 = ax4.bar(travel_attr['BusinessTravel'],
travel_attr['AttritionRate'],
color=t_colors, edgecolor='none',
width=0.5)
ax4.set_ylabel('Attrition Rate (%)', color=muted)
ax4.set_title('Attrition by Travel Frequency', color=navy,
fontweight='bold', pad=10,
fontfamily='serif')
ax4.tick_params(colors=muted)
for spine in ax4.spines.values():
spine.set_color(border)
ax4.grid(color=border, linestyle='--',
linewidth=0.5, axis='y')
ax4.set_xticklabels(travel_attr['BusinessTravel'],
rotation=15, ha='right',
color=muted, fontsize=8)
for bar, val in zip(bars4, travel_attr['AttritionRate']):
ax4.text(bar.get_x() + bar.get_width()/2,
bar.get_height() + 0.2,
f'{val}%', ha='center',
color=navy, fontsize=10,
fontweight='bold')
plt.tight_layout()
plt.show()
```
### Demographic Findings and Recommendations
Younger employees, single employees, and frequent travellers
all show elevated attrition rates. These groups share
a common characteristic, lower organisational attachment
and higher mobility.
**Recommendations:**
- Create a young professional programme targeting
employees Under 25 and 25 to 34, structured
mentorship, visible career paths, and social cohesion
- Review the business travel policy for frequent
travellers, consider travel allowances,
flexible recovery days, and reduced consecutive
travel weeks
- Design retention initiatives sensitive to the
needs of single employees who may prioritise
flexibility and social connection at work
:::
---
# Power BI Dashboard
## Dashboard Overview
The analytical findings above were translated into an interactive
five-page Power BI dashboard for executive and operational use
at Meridian Group. The dashboard enables real-time filtering
by department, job role, gender, and attrition status.
### Page 1 Executive Overview
<div class="dashboard-screenshot">
<img src="screenshots/page1_overview.png" alt="Power BI Dashboard Executive Overview"/>
<div class="dashboard-label">Dashboard Page 1 Executive Overview · Meridian Group People Analytics</div>
</div>
### Page 2 Department and Role Analysis
<div class="dashboard-screenshot">
<img src="screenshots/page2_department.png" alt="Power BI Dashboard Department and Role Analysis"/>
<div class="dashboard-label">Dashboard Page 2, Department and Role Analysis</div>
</div>
### Page 3 Compensation Analysis
<div class="dashboard-screenshot">
<img src="screenshots/page3_compensation.png" alt="Power BI Dashboard Compensation Analysis"/>
<div class="dashboard-label">Dashboard Page 3 Compensation Analysis</div>
</div>
### Page 4 Satisfaction and Wellbeing
<div class="dashboard-screenshot">
<img src="screenshots/page4_satisfaction.png" alt="Power BI Dashboard Satisfaction and Wellbeing"/>
<div class="dashboard-label">Dashboard Page 4 Satisfaction and Wellbeing</div>
</div>
### Page 5 Demographics and Tenure
<div class="dashboard-screenshot">
<img src="screenshots/page5_demographics.png" alt="Power BI Dashboard Demographics and Tenure"/>
<div class="dashboard-label">Dashboard Page 5 Demographics and Tenure</div>
</div>
---
# Conclusions
## Summary of Findings
| Theme | Key Finding | Priority |
|-------|-------------|----------|
| Overall Attrition | 16.1% above industry benchmark | Critical |
| Department and Role | Sales and Level 1 employees most at risk | High |
| Compensation | Leavers earn significantly less than stayers | High |
| Satisfaction | Poor work-life balance is the strongest predictor | High |
| Overtime | Overtime workers leave at 3x the rate of non-overtime workers | Critical |
| Tenure | First two years are the highest-risk attrition window | High |
| Demographics | Young, single, and frequent-traveller employees most at risk | Medium |
## The Three Most Urgent Actions
**1. Address Overtime Immediately**
Overtime is the single most powerful attrition predictor
in this dataset. Employees who work overtime leave at
more than three times the rate of those who do not.
This is not a minor HR issue, it is a structural risk
that requires an operational response within 30 days.
**2. Fix Compensation at the Lower End**
Employees earning under $3,000 per month are leaving
at a disproportionately high rate. A targeted compensation
review for this band will have an immediate impact on
the overall attrition rate.
**3. Protect the First Two Years**
The highest-risk attrition window is the first 24 months.
A structured early-tenure programme, mentorship,
promotion clarity, and regular check-ins, will
significantly reduce the volume of early leavers.
## Data Limitations
- Dataset contains 1,470 employees, sufficient for
pattern identification but limited for subgroup analysis
in smaller departments
- Satisfaction scores are self-reported and subject
to response bias
- No time dimension is available, attrition trends
over time cannot be assessed
- Salary data is in USD, Meridian Group operates
in Manchester, UK, so currency conversion
assumptions apply
## Final Note to Dr. Amara Osei
The data is clear and the diagnosis is unambiguous.
Meridian Group's attrition problem is not random,
it is concentrated, predictable, and therefore preventable.
Overtime, low compensation, poor work-life balance,
and inadequate early-career support are the four pillars
of the attrition problem. Address all four with the
urgency they deserve and the 16.1% rate will fall
to benchmark levels within 18 months.
The intelligence is here. The decisions are yours.
---
**Report prepared by:** Freda Erinmwingbovo, Data Analyst
**Prepared for:** Dr. Amara Osei, CEO Meridian Group
**Date:** February 2026
**Tools:** Python · pandas · matplotlib · Power BI · Quarto
**Dataset:** IBM HR Analytics Employee Attrition Dataset · 1,470 employees · 35 attributes
*This report was produced to professional data science standards.
All findings are reproducible from the code cells above.
No results have been overstated or adjusted to appear more
favourable than the data supports.*
