Back to Portfolio
View Code & Data on GitHub

HR Analytics & EDA: Employee Workforce Report

1,020 employee records, cleaned and analyzed to see whether pay actually lines up with performance.

Table of Contents

Hint: Tap on any chart in this report to expand it full screen.

Project Overview

This project started with a real-world messy employee dataset pulled straight out of an HR system: inconsistent types, missing values, useless columns, the works.

The goal was simple: clean it, understand it, and pull out insights that actually matter.

No fancy ML models, no over-engineering. Just solid data cleaning, thoughtful exploratory analysis, and clear visualizations that speak for themselves.

What was done:

  1. Identified and handled missing values using statistically sound imputation (grouped medians, not blind averages)
  2. Fixed data types: dates stored as strings, salary as floats, age with decimal noise
  3. Dropped irrelevant columns that added zero analytical value
  4. Ran targeted EDA focusing on salary distribution, performance-salary relationships, and department-level patterns
  5. Built visualizations that surface the story behind the numbers

Dataset Summary

Attribute Detail
Total Records 1,020 employees
Columns (Raw) 12
Columns (Final) 10 (after dropping Email & Phone)
Employee ID Range EMP1000 – EMP2019
Unique Departments 36 department-region combinations
Employee Statuses Active, Inactive, Pending
Performance Tiers Excellent, Good, Average, Poor
Salary Range $50,047 – $119,971
Age Range 25 – 40 years
Hire Date Coverage 2020 – 2024

Final clean columns: Employee_ID, First_Name, Last_Name, Age, Department_Region, Status, Join_Date, Salary, Performance_Score, Remote_Work

Data Quality Issues & Cleaning

The raw dataset had several problems that needed attention before any analysis could happen.

Issues Found

Column Issue Records Affected
Age Missing values (NaN) 211 out of 1,020 (~20.7%)
Salary Missing values (NaN) 24 out of 1,020 (~2.4%)
Join_Date Stored as string, not datetime All 1,020 records
Age Float type instead of integer All non-null records
Salary Float with unnecessary decimals All non-null records
Email Not useful for analysis Dropped entirely
Phone Negative/corrupted values, not useful Dropped entirely

How It Was Handled

Age: Missing values were filled using the median age grouped by department-region. Not a flat median across the board, but grouped so that if DevOps-California skews younger and Finance-Texas skews older, the imputation respects that. Histogram and box plot were checked first to confirm the distribution before choosing median over mean.

Salary: Same approach. Grouped by department-region, then filled with the group median. This matters because a missing salary in HR-Florida shouldn't be filled with the overall company average when that department might pay differently.

Join_Date: Converted from raw string format ("4/2/2021") to proper pandas datetime for any time-based analysis.

Email & Phone: Both dropped. Email had no analytical value. Phone numbers were corrupted (negative values like -1651623197), clearly bad data, so they were removed rather than guessed at.

After cleaning: zero null values across all 10 columns. Clean dataset, no compromises.

Exploratory Data Analysis

Salary Distribution Overview

Metric Value
Mean Salary $85,172
Median Salary $85,547
Min Salary $50,047
Max Salary $119,971
Standard Deviation $19,873

The mean and median are nearly identical ($85,172 vs $85,547), which tells us the salary distribution is close to symmetric: no heavy skew toward high or low earners. This is a fairly balanced payroll.

Above vs Below Average Salary

The workforce was split based on the average salary threshold of $85,172.

Category Employee Count Share
Above Average 517 50.7%
Below Average 503 49.3%

Almost a perfect 50/50 split. This further confirms the balanced distribution: the company isn't top-heavy or bottom-heavy in compensation.

Above vs Below Average Salary

Top departments with above-average earners:

Department-Region Count
DevOps-Florida 21
Admin-California 21
Admin-Illinois 21
Sales-Florida 21
Finance-California 20
Cloud Tech-Florida 20

Top departments with below-average earners:

Department-Region Count
Finance-Illinois 24
HR-Florida 22
Admin-Nevada 20
Sales-Nevada 20
DevOps-California 19

A few patterns here. Florida-based departments show up on both sides, which suggests salary variance within the state rather than a uniform pay structure. Finance-Illinois stands out as having the highest concentration of below-average earners.

Age breakdown of above-average earners:

Age Count
30 156
35 129
40 120
25 99
32 13

Employees aged 30 dominate the above-average salary bracket. The 25-year-old group has the smallest representation up top, which makes sense, as they're likely early in their careers.

Top 7 Highest Paid Employees

Rank Name Salary Age Department Status Performance
1 Charlie Smith $119,971 30 Cloud Tech-Texas Inactive Average
2 Grace Smith $119,890 30 HR-Nevada Active Average
3 Bob Williams $119,801 30 Sales-California Active Average
4 Frank Johnson $119,764 40 Sales-Illinois Inactive Average
5 Grace Brown $119,586 30 DevOps-Illinois Active Average
6 Alice Brown $119,574 30 Admin-Illinois Inactive Average
7 Alice Miller $119,407 30 Sales-Florida Inactive Excellent

Top 7 High Salary Earners

Something worth noting that 6 out of the top 7 earners have an "Average" performance score, not "Excellent." And 5 of them are 30 years old. The highest paid employee in the company (Charlie Smith, $119,971) is rated Average and currently Inactive. That raises questions about whether compensation is truly tied to performance, or if there are other factors driving pay at the top.

Top 7 Lowest Paid Employees

Rank Name Salary Age Department Status Performance
1 Heidi Williams $50,047 25 Cloud Tech-New York Inactive Good
2 Heidi Johnson $50,060 25 Finance-New York Pending Poor
3 Charlie Garcia $50,110 35 Admin-Nevada Inactive Average
4 Frank Davis $50,153 40 Cloud Tech-Texas Pending Good
5 Eva Williams $50,173 35 Finance-Florida Inactive Poor
6 Alice Garcia $50,288 35 Sales-New York Active Excellent
7 Grace Smith $50,300 40 Admin-Florida Inactive Good

Top 7 Low Salary Earners

Interestingly, the lowest earners aren't all young employees. Ages range from 25 to 40. Alice Garcia (#6) has an Excellent performance score but earns just $50,288. She's one of the lowest paid people in the entire company despite being a top performer. That's a retention risk flag right there.

Salary vs Performance Deep Dive

This is where it gets interesting. The dataset has 216 employees rated "Poor" out of 1,020 total. That's roughly 21% of the workforce.

The analysis below breaks down salary extremes within performance tiers to answer a simple question: does pay match performance here?

High Earners with Excellent Performance

These are the employees doing it right: top performance, top pay.

Rank Name Salary Age Department Status Remote
1 Alice Miller $119,407 30 Sales-Florida Inactive Yes
2 Alice Williams $118,907 35 Admin-California Active Yes
3 Bob Miller $118,869 35 DevOps-Illinois Active Yes
4 Charlie Brown $118,677 30 HR-New York Active No
5 Frank Smith $118,456 30 HR-Nevada Inactive Yes
6 Eva Jones $118,030 32 Admin-Florida Inactive Yes
7 David Jones $117,870 35 DevOps-Texas Inactive No

Top 7 High Earners - Excellent Performance

Salaries range from $117,870 to $119,407. These are people earning near the company maximum and delivering excellent results. Worth noting that 5 out of 7 are remote workers, and ages are clustered between 30–35. However, 4 out of 7 are Inactive, which is a concern. If your best-performing, highest-paid employees are leaving, that's a serious business problem.

High Earners with Poor Performance

This is the red flag group. High salary, low output.

Rank Name Salary Age Department Status Remote
1 Charlie Miller $119,389 35 Cloud Tech-Florida Active No
2 Alice Smith $119,311 35 Admin-California Pending Yes
3 David Johnson $119,217 30 Cloud Tech-Texas Inactive No
4 Bob Brown $119,152 40 Admin-Florida Pending Yes
5 Frank Jones $118,907 40 Sales-California Inactive No
6 Alice Williams $118,672 40 Finance-Texas Pending No
7 Bob Garcia $118,413 40 Finance-Nevada Inactive No

Top 7 High Earners - Poor Performance

These 7 employees earn between $118,413 and $119,389: nearly the same range as the excellent performers. Charlie Miller earns $119,389 with a Poor rating; Alice Miller (Excellent) earns $119,407. That's an $18 difference for vastly different performance levels.

Also notable: 5 out of 7 are not working remotely, and 4 out of 7 are aged 40 (the oldest bracket in the dataset). This suggests the compensation structure may be rewarding tenure or seniority rather than actual output.

Low Earners with Excellent Performance

These are the undervalued employees: delivering excellent work but sitting at the very bottom of the pay scale.

Rank Name Salary Age Department Status Remote
1 Alice Garcia $50,288 35 Sales-New York Active No
2 Frank Davis $50,593 30 HR-Illinois Active Yes
3 David Miller $51,054 25 Admin-Nevada Inactive No
4 Frank Brown $51,074 25 Finance-Texas Pending Yes
5 Heidi Garcia $51,150 25 Cloud Tech-Florida Pending No
6 Frank Smith $51,311 25 Finance-Nevada Active No
7 David Smith $51,939 40 Admin-Texas Active Yes

Top 7 Low Earners - Excellent Performance

These employees earn between $50,288 and $51,939 while consistently rated Excellent. Alice Garcia brings in $50,288. Compare that to Charlie Miller (Poor) earning $119,389. That's a $69,101 gap between a top performer and a bottom performer.

4 out of 7 here are 25 years old, which might explain the lower salary as entry-level positioning, but performance ratings don't care about age. If someone's delivering Excellent results, they should be compensated accordingly. This group represents the highest retention risk in the organization.

Low Earners with Poor Performance

Rank Name Salary Age Department Status Remote
1 Heidi Johnson $50,060 25 Finance-New York Pending No
2 Eva Williams $50,173 35 Finance-Florida Inactive No
3 Heidi Johnson $50,651 30 Admin-Nevada Pending No
4 Charlie Jones $51,533 35 DevOps-New York Inactive Yes
5 David Smith $51,673 32 Admin-New York Active No
6 Grace Davis $51,865 30 HR-Nevada Pending Yes
7 Alice Miller $52,289 25 Finance-Nevada Inactive Yes

Top 7 Low Earners - Poor Performance

Low pay, low performance. This group is where it lines up: salary reflects output. The concerning pattern is that 5 out of 7 are not remote workers, and the statuses lean heavily toward Inactive/Pending rather than Active. These may be employees who were already on their way out.

Key Business Takeaways

1. Compensation Is Not Aligned with Performance

The most critical finding. Employees rated "Poor" earn almost identical salaries to those rated "Excellent" at the top end (~$119K for both). The pay-for-performance model appears broken, or it doesn't exist at all.

2. High-Performing, Low-Paid Employees Are a Retention Risk

7 employees rated Excellent earn between $50K–$52K. These are people delivering top results at bottom-tier pay. If compensation isn't corrected, the company risks losing its best talent, especially Alice Garcia ($50,288, Excellent, Active).

3. 21% of the Workforce Is Rated "Poor"

216 out of 1,020 employees carry a Poor performance score. That's over one-fifth of the workforce. Combined with the fact that some of these employees are among the highest paid, this suggests a need for performance-based compensation reviews.

4. Age 30 Dominates the High-Earning Bracket

The 30-year-old cohort leads across above-average earners (156 employees). This could indicate a sweet spot where employees have enough experience to command higher salaries but haven't yet hit career plateaus.

5. Remote Work Correlates with Higher Performance

Among high earners with Excellent performance, 5 out of 7 work remotely. Among high earners with Poor performance, only 2 out of 7 are remote. While this isn't causal, it's a pattern worth investigating further.

6. Florida and California Show Up Everywhere

These two states appear across both the highest and lowest salary brackets, the best and worst performers. They likely represent the company's largest regional hubs and may need localized compensation strategies.

7. Inactive Status Among Top Performers Is Alarming

4 out of 7 highest-paid Excellent performers are Inactive. If the company's best people are leaving despite good pay, the problem goes beyond compensation; it could be culture, growth opportunities, or management.

Tools & Technologies

Tool Purpose
Python Core language for data processing and analysis
Pandas Data cleaning, manipulation, groupby operations, imputation
Matplotlib All visualizations: bar charts, horizontal bars, histograms, box plots
Jupyter Notebook Interactive development environment for the full analysis pipeline

Analysis conducted on a real-world messy dataset. All values reflect the cleaned and processed data.

Back to Portfolio