California Unemployment Rate Analysis (2008–2026)

A descriptive and exploratory analysis of California unemployment rates across seven age groups, examining differences in unemployment levels, variability, temporal behavior, historical extremes, calendar-month patterns, and statistical co-movement.

Project Overview

Unemployment does not affect all age groups in the same way. Differences may appear in typical unemployment levels, the magnitude of fluctuations, the timing of historical highs and lows, and the extent to which age groups move together over time.

This project examines California unemployment rates across seven age groups from February 2008 through June 2026.

The analysis is intentionally descriptive and exploratory. Its purpose is to characterize patterns present in the available data rather than establish causal relationships, make predictions, or perform statistical inference.

The project follows a reproducible analytical workflow from data acquisition and quality assessment through exploratory data analysis and final interpretation.

Analytical Objectives

The project was designed to:

• Describe typical unemployment-rate levels and distributions across age groups.
• Compare unemployment-rate variability in absolute and relative terms.
• Examine how unemployment rates changed over the period represented by the dataset.
• Identify historical high and low unemployment rates and substantial year-over-year changes.
• Explore recurring differences across calendar months.
• Evaluate statistical co-movement among age groups in both unemployment-rate levels and year-over-year changes.
• Identify patterns that may warrant future investigation.

Key Findings

Age-group differences are substantial

Ages 16–19 have the highest average unemployment rate at approximately 22.93%, followed by ages 20–24 at approximately 11.71%.

Average unemployment generally declines across the younger age groups before becoming more similar among the middle and older groups. The pattern is not completely monotonic, with a slight increase in average unemployment among the oldest groups.

Variability depends on how it is measured

Ages 16–19 exhibit the greatest absolute variability, with a standard deviation of approximately 7.36 percentage points.

However, relative variability produces a different result. Using the coefficient of variation, ages 45–54 exhibit the greatest variability relative to their average unemployment level.

This distinction demonstrates why analytical conclusions depend on clearly defining the metric being evaluated.

Age groups move together, but not identically

Unemployment rates across the seven age groups exhibit substantial shared temporal movement. Periods of increasing and decreasing unemployment frequently occur during similar portions of the time series.

The groups nevertheless differ in unemployment levels, magnitude of movement, and the exact timing of historical maxima and minima.

Statistical co-movement is substantial

Pairwise correlations in unemployment-rate levels range from approximately 0.751 to 0.990, while correlations in year-over-year changes range from approximately 0.722 to 0.966.

All age-group pairs are positively correlated, but co-movement in year-over-year changes is consistently weaker than co-movement in unemployment-rate levels.

These relationships are descriptive statistical associations and should not be interpreted as evidence of causation.

Additional patterns warrant future investigation

Exploratory analysis identified modest differences in average unemployment rates across calendar months, including a common pattern in which October has the highest monthly average across all seven age groups.

An exploratory period comparison also found stronger year-over-year correlations for most age-group pairs from 2020 onward.

These observations identify potential directions for future analysis but do not establish formal seasonality, structural change, or causal explanations.

Dataset

The project uses the Unemployment Rate by Age Groups dataset published through California public data resources.

Source organization: California Employment Development Department Geographic coverage: California Temporal coverage: February 2008 – June 2026 Observation frequency: Monthly Age groups:

• 16–19
• 20–24
• 25–34
• 35–44
• 45–54
• 55–64
• 65+

The source dataset contains one known temporal gap: October 2025 is absent from the original source data. The missing observation was preserved rather than imputed or interpolated.

The original raw dataset is retained unchanged in the repository.

Analytical Approach

The project follows a multi-stage analytical workflow:

1. Problem Definition

The analytical problem, objectives, scope, boundaries, and analytical questions are established before analysis begins.

2. Data Acquisition

The source dataset, provenance, field definitions, temporal coverage, and known source limitations are documented.

3. Data Profiling and Preparation

The dataset is evaluated for:

• schema and data types;
• missing values;
• duplicate records;
• categorical consistency;
• temporal consistency;
• geographic consistency;
• expected value ranges; and
• source-data gaps.

An analysis-ready dataset is then created while preserving the original raw data.

4. Exploratory Data Analysis

The analysis examines:

• descriptive statistics;
• measures of central tendency;
• absolute and relative dispersion;
• distribution shape and skewness;
• IQR-based outlier screening;
• age-group comparisons;
• time-series behavior;
• calendar-aware year-over-year changes;
• calendar-month averages;
• historical maxima and minima;
• pairwise correlations;
• period-specific correlations; and
• selected pairwise relationships.

5. Findings and Conclusion

Evidence developed during EDA is synthesized to answer the project's analytical questions without introducing new analytical methods.

Repository Structure

california-unemployment-analysis/
│
├── README.md
│
├── notebooks/
│   ├── 01_Introduction.ipynb
│   ├── 02_Data_Acquisition.ipynb
│   ├── 03_Data_Profiling_and_Preparation.ipynb
│   ├── 04_Exploratory_Data_Analysis.ipynb
│   └── 05_Findings_and_Conclusion.ipynb
│
├── data/
│   ├── raw/
│   │   └── unemployment_rate_by_age_groups.csv
│   │
│   └── processed/
│       ├── california_unemployment_analysis_ready.csv
│       └── california_unemployment_analysis_ready.parquet
│
└── docs/

Notebook Guide

Notebook	Purpose
01 — Introduction	Defines the problem statement, analytical objectives, scope, boundaries, and analytical questions.
02 — Data Acquisition	Documents the dataset, source provenance, field definitions, coverage, and known source limitations.
03 — Data Profiling & Preparation	Profiles data quality, prepares the analysis-ready dataset, validates transformations, and exports processed data.
04 — Exploratory Data Analysis	Performs the descriptive, comparative, temporal, and relationship analyses used to develop the project's evidence.
05 — Findings & Conclusion	Synthesizes the evidence, answers the analytical questions, documents limitations, and identifies opportunities for future analysis.

For a complete walkthrough, begin with 01 — Introduction and proceed through the notebooks sequentially.

Tools and Technologies

The project uses:

• Python
• Jupyter Notebook
• Pandas
• Matplotlib
• CSV
• Apache Parquet

The workflow emphasizes reproducibility, separation of raw and processed data, explicit validation, and clear documentation of analytical decisions.

Skills Demonstrated

This project demonstrates practical application of:

• analytical problem formulation;
• analytical-question development;
• data acquisition and provenance documentation;
• data profiling and quality assessment;
• data preparation and validation;
• descriptive statistics;
• measures of central tendency and dispersion;
• distribution analysis;
• exploratory data analysis;
• time-series exploration;
• year-over-year change analysis;
• correlation analysis;
• data visualization;
• statistical interpretation;
• analytical reasoning;
• reproducible workflow design;
• limitation identification; and
• communication of analytical findings.

Particular attention is given to distinguishing concepts that can easily be conflated, including absolute versus relative variability, unemployment levels versus changes, co-movement versus causation, and exploratory calendar-month patterns versus formally established seasonality.

Limitations

The dataset contains statewide unemployment rates by age group but does not include additional economic, demographic, geographic, industry, policy, or labor-market variables that could explain the observed patterns.

The dataset begins in February 2008 and ends in June 2026, and October 2025 is absent from the source data.

The analysis is limited to descriptive and exploratory methods. It does not include:

• inferential statistical testing;
• causal analysis;
• predictive modeling;
• forecasting; or
• formal time-series modeling.

Correlations and temporal co-movement therefore describe statistical relationships within the observed data and should not be interpreted as causal relationships.

Opportunities for Future Analysis

The findings establish several potential directions for future projects, including incorporating explanatory economic or demographic variables, applying statistical inference, formally investigating seasonal and time-series structure, examining changes in co-movement across periods, extending the analysis to more detailed geographic or population data, and eventually developing forecasting models.

These are treated as separate future analytical objectives, not unfinished components of the current project.

Project Scope

This project represents a completed descriptive and exploratory data-analysis workflow.

Its purpose is not to explain why California unemployment changed, predict future unemployment, or establish causal relationships. Instead, it establishes a reproducible empirical description of how unemployment rates differ across age groups, vary over time, and move in relation to one another within the available data.

Author

Tyler V. McDougal

Data analysis portfolio project 2026