Python for HR Analytics Tutor

Specialized skill for HR analytics Python work. General teaching preferences (explain WHY, beginner-friendly, error handling, etc.) are defined in the global claude.md — this skill adds HR-specific context and patterns.

When to Load References

Load these files based on the task:

• references/python-basics.md — For foundational Python concepts (data types, variables, DataFrames, basic operators, errors). Use when revisiting fundamentals.

• references/hr-patterns.md — For specific HR analytics operations (turnover calculations, time-to-fill, cohort analysis, data merging, aggregations). Use for "how do I do X" questions.

• references/complete-examples.md — For end-to-end workflows (comprehensive turnover analysis, recruiting funnel, retention cohorts, forecasting). Use for understanding full analytical processes.

Loading strategy:

• Basic concept questions → python-basics.md
• Specific pandas/HR operations → hr-patterns.md
• Complete workflow examples → complete-examples.md
• Complex projects → may need multiple references

HR Analytics Domain Context

Common Metrics

• Turnover rate: (terminations / headcount) × 100 — voluntary, involuntary, by segment
• Time-to-fill: Days from req open to offer accept — by role, office, aged reqs
• Retention rate: Cohort-based (% still employed after X months) or milestone-based
• Headcount: Point-in-time counts, trending by month/segment
• Span of control: Direct reports per manager

Common Data Operations

• Merging employee data across systems (HRIS, ATS, payroll)
• Fuzzy name matching across sources
• Date range filtering (active during period, hired between dates)
• Aggregating by multiple dimensions (role + office + tenure band)
• Creating derived fields (tenure buckets, age bands, performance tiers)

Analysis Types

• Cohort analysis: Retention by hire period
• Funnel analysis: Recruiting stages and conversion rates
• Trend analysis: Metrics over time
• Segment comparison: By role, office, tenure, performance
• Forecasting: Headcount projections, attrition predictions

HR-Specific Code Patterns

Turnover Rate by Segment

# Count terminations per segment
terms = df[df['is_terminated']].groupby('segment').size()

# Count total headcount per segment
headcount = df.groupby('segment').size()

# Calculate rate
turnover_rate = (terms / headcount * 100).round(2)

Time-to-Fill Calculation

# Convert dates and calculate days
df['req_open'] = pd.to_datetime(df['req_open'])
df['offer_accept'] = pd.to_datetime(df['offer_accept'])
df['time_to_fill'] = (df['offer_accept'] - df['req_open']).dt.days

# Average by role
ttf_by_role = df.groupby('role')['time_to_fill'].mean().round(1)

Tenure Calculation

# Calculate tenure in years (accounting for leap years)
df['tenure_years'] = (pd.Timestamp.now() - df['hire_date']).dt.days / 365.25

# Create tenure bands
df['tenure_band'] = pd.cut(
    df['tenure_years'],
    bins=[0, 1, 3, 5, 10, float('inf')],
    labels=['<1 yr', '1-3 yrs', '3-5 yrs', '5-10 yrs', '10+ yrs']
)

Cohort Retention

# Create hire cohort (year-month)
df['hire_cohort'] = df['hire_date'].dt.to_period('M')

# Calculate months since hire
df['months_since_hire'] = (
    (pd.Timestamp.now() - df['hire_date']).dt.days / 30.44
).astype(int)

# Retention at milestone
milestone = 12  # months
cohort_size = df.groupby('hire_cohort').size()
still_active = df[
    (df['is_active']) & (df['months_since_hire'] >= milestone)
].groupby('hire_cohort').size()
retention_rate = (still_active / cohort_size * 100).round(1)

Recruiting Funnel

# Stage conversion rates
funnel = df.groupby('stage').size().reset_index(name='count')
funnel['conversion'] = (
    funnel['count'] / funnel['count'].iloc[0] * 100
).round(1)

# Stage-to-stage drop-off
funnel['stage_conversion'] = (
    funnel['count'] / funnel['count'].shift(1) * 100
).round(1)

Data Source Context

Typical data sources in this environment:

• BigQuery: Primary data warehouse
• Workday RaaS: HR system reports
• CSV/Excel exports: Ad-hoc data pulls

Common data quality issues to watch for:

• Multiple records per employee (need to deduplicate or filter to latest)
• Date fields as strings (need pd.to_datetime conversion)
• Inconsistent department/role naming across systems
• Missing termination dates for active employees (expected, not an error)