Seasonality Adjustment Engine

python
# Seasonal Adjustment Framework
monthly_seasonal_factors = {
    'January': 1.12,    # Deductible reset, elective procedures scheduled
    'February': 1.08,   # Continued deductible optimization
    'March': 1.05,      # Deductible satisfied, utilization normalizing
    'April': 0.98,      # Spring lull
    'May': 0.95,        # Pre-summer drop
    'June': 0.94,       # Summer vacation begins
    'July': 0.92,       # Peak vacation season
    'August': 0.91,     # Continued summer lull
    'September': 1.02,  # Back-to-school checkups
    'October': 1.04,    # Flu season begins
    'November': 0.97,   # Holiday deferrals begin
    'December': 0.93    # Year-end holidays, elective procedures deferred
}

def seasonally_adjust_costs(monthly_claims):
    """
    Remove seasonal patterns to expose true underlying trend
    """
    adjusted_claims = []
    
    for month_data in monthly_claims:
        month_name = month_data.month
        actual_pmpm = month_data.total_claims / month_data.member_months
        
        # Remove seasonal effect
        seasonal_factor = monthly_seasonal_factors[month_name]
        adjusted_pmpm = actual_pmpm / seasonal_factor
        
        adjusted_claims.append({
            'month': month_name,
            'actual_pmpm': actual_pmpm,
            'seasonal_factor': seasonal_factor,
            'seasonally_adjusted_pmpm': adjusted_pmpm
        })
    
    return adjusted_claims

def calculate_deseasonalized_trend(current_period, prior_period):
    """
    Compare same months year-over-year OR use seasonally-adjusted values
    """
    # Method 1: Year-over-year (natural deseasonalization)
    yoy_trend = (current_period.pmpm / prior_period.pmpm) - 1
    
    # Method 2: Seasonally-adjusted month-to-month
    current_adjusted = current_period.pmpm / monthly_seasonal_factors[current_period.month]
    prior_adjusted = prior_period.pmpm / monthly_seasonal_factors[prior_period.month]
    mom_trend = (current_adjusted / prior_adjusted) - 1
    
    return {
        'yoy_trend': yoy_trend,
        'mom_seasonally_adjusted': mom_trend
    }

# Example: Q1 vs. Q3 Comparison
# Q1 Actual: $1,085 PMPM (Jan 1.12x, Feb 1.08x, Mar 1.05x avg = 1.083x)
# Q1 Adjusted: $1,002 ($1,085 / 1.083)
#
# Q3 Actual: $945 PMPM (Jul 0.92x, Aug 0.91x, Sep 1.02x avg = 0.950x)
# Q3 Adjusted: $995 ($945 / 0.950)
#
# Without adjustment: looks like 14.8% decline (panic!)
# With adjustment: actually 0.7% improvement (normal variance)