Portfolio Project · EDA Report · 2024
AI Job Replacement
& Skill Shift Analysis
A rigorous exploratory data analysis investigating how artificial intelligence is reshaping employment risk, salary outcomes and workforce skill demands across 15,000 records, 8 industries, 9 countries and 7 years of data from 2020 to 2026.
KS
Kevin Savaliya
Data Analyst · EDA Portfolio
15,000 Records20 Features8 Industries9 Countries2020–20260 Missing Values
15,000
Total Records
Job records analysed
20
Features
Variables per record
73%
At-Risk Jobs
Medium or high automation exposure
r=0.96
Top Correlation
Automation risk vs AI replacement
~0%
Salary Change
Avg delta post-AI adoption
60.74%
Highest Risk
Truck Driver automation risk
34.85%
Lowest Risk
Software Engineer automation risk
Stable
7-Year Trend
All AI metrics flat 2020-2026
Dataset Overview
The AI Job Replacement Dataset (Kaggle) — 15,000 rows × 20 columns, zero missing values, zero duplicates.
15,000
Records
20
Features
14
Numeric
4
Categorical
8
Industries
9
Countries
10
Roles
7 yrs
2020–2026
Key Variables
automation_risk_percentai_replacement_scoresalary_before_usdsalary_after_usdreskilling_urgency_scoreskill_gap_indexai_adoption_levelremote_feasibility_score
Key Findings
73%
Of all jobs face medium-to-high automation risk — a present structural reality across every industry and country.
r = 0.96
Pearson correlation between automation_risk_percent and ai_replacement_score — the single dominant signal in the dataset.
~0%
Average salary change after AI adoption across all 15,000 records. AI disrupts skills before wages.
Stable
All AI metrics (risk, replacement, reskilling) are flat across 7 years — gradual, not accelerating.
60.74%
Truck Driver automation risk — highest of any role. Software Engineers sit safest at 34.85%.
1.4 pts
Max industry risk spread (45.60%–47.02%). Job role predicts risk far better than industry.
Analysis Sections — Click to Explore
Methodology & Tools
PANDAS
Data loading, cleaning, groupby aggregations and feature engineering
PLOTLY
Interactive charts — bar, sunburst, sankey, radar, choropleth, treemap
SEABORN
Correlation matrices, pivot heatmaps and distribution histograms
SCIPY
Pearson correlation, statistical significance and numerical operations
JUPYTER
Interactive analysis environment — 68 cells executed end-to-end
KAGGLE
AI Job Replacement Dataset — public, structured, 0 missing values
Univariate Analysis
Individual feature distributions, counts and risk breakdowns across all 15,000 records.
Analysis By Kevin Savaliya
Feature Distributions — All Numeric Variables
INSIGHTNear-uniform distributions for automation_risk_percent and ai_replacement_score. Salary bell curves center at ~$89,771. wage_volatility_index is most right-skewed.
Top Job Roles by Record Count
INSIGHTMarketing Specialist leads (1,551 / 10.34%), Accountant lowest (1,415 / 9.43%). Stratified design ensures unbiased cross-role comparisons.
Top Industries by Record Count
INSIGHTFinance leads at 1,942 (12.95%), Education 1,814 (12.09%). All 8 industries are well-balanced — findings reflect genuine AI impact differences.
Top Countries by Record Count
INSIGHTAll 9 countries contribute ~1,580-1,720 records each. Globally balanced coverage enables fair cross-country comparisons.
Top High-Paying Roles After AI Adoption
INSIGHTAll 10 roles cluster between $88,256–$91,376 — only a ~$3K spread. AI has not created meaningful pay differentiation between job types.
Automation Risk Category Breakdown
INSIGHTMedium: 43.1% (6,464) · High: 30.0% (4,496) · Low: 26.9% (4,040). Over 73% of all jobs face medium-to-high automation risk.
Risk Category Distribution
INSIGHTMedium (43.1%) is the largest segment. High (30.0%) outpaces Low (26.9%). AI displacement is pervasive across the full workforce.
Bivariate Analysis
Pairwise relationships — how automation risk connects to AI scores, salaries and reskilling urgency.
Analysis By Kevin Savaliya
Industry Automation Risk — Sunburst
INSIGHTEvery industry shows a similar Low/Medium/High tri-layer. Transportation and Manufacturing carry slightly thicker High-risk segments. Job role is a stronger predictor than industry.
Salary Change After AI Adoption by Role
INSIGHTSalary deltas are near-zero for all roles — largest change only ±$200 on a ~$90K base. AI disrupts skills before wages.
Average AI Impact — 4 Metrics by Industry
INSIGHTskill_gap_index (~49–51) dominates in scale, then automation_risk (~45–47), reskilling_urgency (~35–36), ai_disruption (~22–24). Every industry shows the same relative ordering.
Correlation Analysis
Statistical co-movement between all 14 numeric features — isolating the dominant automation vs replacement signal.
Analysis By Kevin Savaliya
Full Feature Correlation Matrix (Pearson r)
INSIGHTOne dominant pair: automation_risk_percent vs ai_replacement_score r = 0.96. All other pairs near-zero. wage_volatility_index correlates with essentially nothing.
Feature Correlation with ai_replacement_score
INSIGHTautomation_risk_percent r = 0.9644. All 13 other features show |r| < 0.05. No other factor independently predicts AI replacement risk.
Categorical Analysis
Comparing job roles, industries and countries across every key AI impact metric.
Analysis By Kevin Savaliya
Automation Risk % by Job Role
INSIGHTTruck Drivers 60.74% and Customer Support Reps 59.85% face highest risk. Software Engineers 34.85% are safest. A 26-point spread is the widest differentiator in the dataset.
Reskilling Urgency Score by Job Role
INSIGHTEven the safest roles (SW Engineers: 30.83) carry significant urgency — no role is exempt from continuous learning.
Automation Risk % by Industry
INSIGHTRange only 45.60%–47.02% — just 1.4 points across all 8 industries. Industry is a weak predictor of automation exposure.
Reskilling Urgency Score by Industry
INSIGHTFlat across all 8 industries (35.41–36.18). Every sector faces the same reskilling pressure.
Average Salary Change % by Job Role
INSIGHTRange: -0.22% to +0.12% — statistically zero. AI is a skill crisis before a wage crisis.
Top Countries by AI Adoption Level
INSIGHTIndia leads at 50.46, Australia (50.34), Singapore (50.01). All countries within the narrow 49.5–50.5 range — globally consistent adoption.
AI Disruption Intensity by Country
INSIGHTRange: only 23.07–23.53 across 9 countries. Disruption follows job role composition, not national AI policy.
Skill Demand Growth % by Job Role
INSIGHTSoftware Engineers 5.59% lead. Even high-risk Truck Drivers (4.71%) show positive demand — retraining pathways exist.
Skill Demand Growth % by Industry
INSIGHTFinance and Technology lead. Healthcare lags (4.60%). Early adopters create max disruption and max opportunity.
AI Talent Demand by Country
INSIGHTAll countries show similar talent demand (~4.8–5.2%). Globally distributed, not concentrated in tech hubs.
Remote Feasibility Score by Industry
INSIGHTTechnology (57.08) leads; Manufacturing (53.66) lowest. WFH correlates ~0 with automation risk — zero protection against displacement.
Multivariate Analysis
Pivot heatmaps and composite charts combining 3+ variables to reveal hidden cross-dimensional patterns.
Analysis By Kevin Savaliya
Salary Change % — Job Role × Country
INSIGHTChanges scatter randomly around zero. HR Manager × India peaks at +1.8%. AI salary impact is country-agnostic.
Skill Gap Index — Country × Industry
INSIGHTBroadly uniform (47–55). Brazil × Retail peaks at 54.9 — widest skill shortage. A global structural issue.
AI Replacement Score — Industry × Year
INSIGHTAll cells hover in 44–50. Energy 2026 reaches 50 — highest cell. Disruption is baked into job structure.
Country AI Adoption vs Automation Risk Scatter
INSIGHTIndia leads AI adoption (50.46) yet has average automation risk (~46%). National AI investment does not equal workforce displacement.
AI Risk Distribution Treemap
INSIGHTNo safe corner — every industry-role combination carries all three risk categories. Medium-risk tiles consistently largest.
Time Series Analysis
Tracking disruption metrics from 2020 to 2026 — trends, stability and salary shifts over time.
Analysis By Kevin Savaliya
Average Key Metrics Trend (2020–2026)
INSIGHTAll three lines essentially flat 2020–2026. AI impact is consistently present but not accelerating.
Salary Change % by Industry over Time
INSIGHTAll 8 industry lines oscillate around zero (-0.84% to +1.31%). No directional wage trend over 7 years.
Risk Category Distribution by Year
INSIGHTRemarkably stable: Medium ~900/yr, High ~640/yr, Low ~570/yr. No dramatic year-on-year spike.
Skill Demand Growth Trend 2020–2026
INSIGHTDemand fluctuates 4.52% (2022) to 5.30% (2026). AI-driven skill demand builds gradually.
Salary Before vs After AI Adoption by Year
INSIGHTBefore and after salary lines are completely overlapping across all 7 years at ~$89K–$90K. Zero measurable wage disruption.
Geographic Analysis
Global mapping of AI disruption intensity across 9 countries worldwide.
Analysis By Kevin Savaliya
Global AI Workforce Disruption Map
INSIGHTDisruption nearly uniform worldwide. Developed and emerging economies show equal exposure. AI displacement is a truly global phenomenon.
Country AI Adoption vs Automation Risk
INSIGHTIndia leads AI adoption (50.46) yet has average automation risk (~46%). National investment does not equal workforce displacement.
AI Disruption Intensity by Country
INSIGHTSpans only 23.07–23.53. Disruption follows job role composition, not national policy.
Advanced Analysis
High-impact synthesis: future-proof roles, cross-country flows and multi-dimensional industry profiling.
Analysis By Kevin Savaliya
Future-Proof Job Roles — Risk vs Skill Demand
INSIGHTSoftware Engineers (35%, 5.59%) and Data Analysts (35%, 5.21%) are clearly future-proof. Truck Drivers (61%) face maximum exposure with lowest skill demand growth.
Sankey: Country → Industry → Risk Category
INSIGHTBalanced flows throughout. Medium risk receives widest bands from all sources, matching the consistent 43% proportion.
Industry Multi-Metric Radar (Normalised 0–1)
INSIGHTEnergy has the largest footprint — high on all 5 axes. Education is most protected. Retail shows disproportionate reskilling urgency relative to its automation risk.