Seasonality Detection

Detect and analyze seasonal patterns in your time series data.

---

anofox_fcst_ts_detect_seasonality

Detect if a series has seasonal patterns at a specific period.

Parameters

Parameter	Type	Required	Default	Description
values	DOUBLE[]	Yes	-	Time series values
period	INTEGER	Yes	-	Period to test

Output

Returns BOOLEAN indicating if seasonality is detected.

Example

SELECT anofox_fcst_ts_detect_seasonality(
    LIST(sales ORDER BY date),
    7  -- period to test
) as is_seasonal
FROM sales_data;

Use when:

• Exploring data
• Deciding on model
• Validating forecast assumptions

---

anofox_fcst_ts_analyze_seasonality

Detailed seasonality analysis with strength metrics and detected periods.

Parameters

Parameter	Type	Required	Default	Description
table_name	VARCHAR	Yes	-	Source table
group_col	VARCHAR	No	NULL	Group column
date_col	VARCHAR	Yes	-	Date column
value_col	VARCHAR	Yes	-	Value column
params	MAP	No	MAP	Configuration

Output

Returns STRUCT containing:

Field	Type	Description
detected_periods	INTEGER[]	All detected seasonal periods
primary_period	INTEGER	Most significant period
seasonal_strength	DOUBLE	Strength of seasonality (0-1)
trend_strength	DOUBLE	Strength of trend (0-1)

Example

SELECT * FROM anofox_fcst_ts_analyze_seasonality(
    'sales_data',
    NULL,  -- group_col
    date,
    sales,
    MAP{}
);

Use when:

• Multiple seasonality suspected
• Hourly/sub-daily data
• Complex patterns

---

Seasonality Analysis Workflow

-- Step 1: Quick check for weekly seasonality
SELECT anofox_fcst_ts_detect_seasonality(
    LIST(sales ORDER BY date),
    7
) as has_weekly_pattern
FROM sales_data;

-- Step 2: Detailed analysis if seasonality detected
SELECT * FROM anofox_fcst_ts_analyze_seasonality(
    'sales_data',
    NULL,
    date,
    sales,
    MAP{}
);

-- Step 3: Use findings to select appropriate model
-- High seasonal_strength → Use AutoETS or HoltWinters
-- Multiple periods → Use AutoTBATS or MSTL

---

Interpretation Guide

Metric	Value	Interpretation
seasonal_strength	> 0.6	Strong seasonality, use seasonal models
seasonal_strength	0.3 - 0.6	Moderate seasonality
seasonal_strength	< 0.3	Weak or no seasonality, try non-seasonal models
trend_strength	> 0.6	Strong trend, consider trend components
trend_strength	< 0.3	Little trend, simpler models may work

---