anofox-forecast-eda

Name: anofox-forecast-eda
Author: DataZooDE

35stars3forks

How to install anofox-forecast-eda?

datazoode/anofox-forecast/anofox-forecast-eda

$npx -y skills add DataZooDE/anofox-forecast --skill anofox-forecast-eda

Installs into the current project.

Summary

Exploratory data analysis and data quality for the anofox_forecast DuckDB extension — 34 per-series statistics, data-quality scoring, quality-report summaries, and 117 tsfresh-compatible feature extraction. Use before forecasting to understand series characteristics (length, gaps

Files · 1

View on GitHub

SKILL.md

1	# Anofox Forecast — EDA & Data Quality Cheat Sheet
2
3	Extension: `anofox_forecast` v0.15.3 \| DuckDB: v1.4.5 LTS / v1.5.4+ \| Dual naming: `ts_` and `anofox_fcst_ts_`
4
5	Understand your data before modelling it: per-series statistics, data-quality scores, feature vectors for ML.
6
7	## Statistics
8
9	### `ts_stats` (table function) — 34 metrics per series
10
11	```sql
12	ts_stats(source VARCHAR, group_col COLUMN, date_col COLUMN, value_col COLUMN,
13	frequency VARCHAR) → TABLE
14	```
15
16	Returns 34 columns: `length`, `n_nulls`, `n_nan`, `n_zeros`, `n_positive`, `n_negative`, `n_unique_values`, `is_constant`, `n_zeros_start`, `n_zeros_end`, `plateau_size`, `plateau_size_nonzero`, `mean`, `median`, `std_dev`, `variance`, `min`, `max`, `range`, `sum`, `skewness`, `kurtosis`, `tail_index`, `bimodality_coef`, `trimmed_mean`, `coef_variation`, `q1`, `q3`, `iqr`, `autocorr_lag1`, `trend_strength`, `seasonality_strength`, `entropy`, `stability`, plus date-derived `expected_length` and `n_gaps`.
17
18	```sql
19	-- Every stat for every series
20	SELECT * FROM ts_stats('sales', product_id, ds, y, '1d');
21
22	-- Quick sanity check
23	SELECT product_id, length, n_nulls, n_gaps, trend_strength, seasonality_strength
24	FROM ts_stats('sales', product_id, ds, y, '1d')
25	WHERE length < 30 OR n_gaps > 0
26	ORDER BY n_gaps DESC;
27	```
28
29	### `ts_stats_agg` (aggregate)
30
31	For custom `GROUP BY` shapes. Takes `(date, value)` — DO NOT wrap in `LIST(...)`.
32
33	```sql
34	ts_stats_agg(date_col TIMESTAMP, value_col DOUBLE)
35	→ STRUCT(length UBIGINT, n_nulls UBIGINT, …, mean DOUBLE, …)
36	```
37
38	```sql
39	SELECT product_id,
40	ts_stats_agg(ds, y).mean AS mean,
41	ts_stats_agg(ds, y).seasonality_strength AS seas_str
42	FROM sales GROUP BY product_id;
43	```
44
45	### `ts_stats_by` — alias for `ts_stats`
46
47	### `ts_stats_summary` — aggregate stats across all groups
48
49	Summarise the per-group stats into panel-level statistics (mean, std, min, max of each metric).
50
51	```sql
52	SELECT * FROM ts_stats_summary('sales', product_id, ds, y, '1d');
53	```
54
55	## Data quality
56
57	### `ts_data_quality` (table function) — per-series score card
58
59	```sql
60	ts_data_quality(source VARCHAR, unique_id_col COLUMN, date_col COLUMN, value_col COLUMN,
61	n_short INTEGER, frequency VARCHAR) → TABLE
62	```
63
64	Returns `unique_id` (group column renamed — the extension normalises it to `unique_id` in this output, even if the input col was `product_id`), plus `overall_score`, `structural_score`, `temporal_score`, `magnitude_score`, `behavioral_score`, `n_gaps`, `n_missing`, `is_constant`.
65
66	`n_short` is the series-length threshold below which a series is flagged short (typical: 14 for daily, 12 for monthly).
67
68	```sql
69	SELECT * FROM ts_data_quality('sales', product_id, ds, y, 14, '1d');
70
71	-- Rank series by quality (note: output col is `unique_id`, not `product_id`)
72	SELECT unique_id, overall_score, structural_score, temporal_score
73	FROM ts_data_quality('sales', product_id, ds, y, 14, '1d')
74	ORDER BY overall_score;
75	```
76
77	### `ts_data_quality_agg` (aggregate)
78
79	Same as `ts_stats_agg` — takes `(date, value)`, returns a STRUCT.
80
81	```sql
82	SELECT product_id,
83	ts_data_quality_agg(ds, y).overall_score AS q
84	FROM sales GROUP BY product_id;
85	```
86
87	### `ts_data_quality_summary` — panel-level roll-up
88
89	```sql
90	SELECT * FROM ts_data_quality_summary('sales', product_id, ds, y, 14);
91	```
92
93	### `ts_quality_report` — human-readable report
94
95	## Feature extraction (117 tsfresh-compatible features)
96
97	### `ts_features_by` — extract all 117 features
98
99	```sql
100	ts_features_by(source VARCHAR, group_col COLUMN, date_col COLUMN, value_col COLUMN) → TABLE
101	```
102
103	Returns the group column + 116 feature columns: `mean`, `standard_deviation`, `skewness`, `kurtosis`, `length`, `linear_trend_slope`, `autocorrelation_lag1`, …
104
105	```sql
106	-- Extract all features
107	SELECT * FROM ts_features_by('sales', product_id, ds, y);
108
109	-- Filter by feature values
110	SELECT product_id, mean, linear_trend_slope
111	FROM ts_features_by('sales', product_id, ds, y)
112	WHERE length > 30 AND autocorrelation_lag1 > 0.5;
113	```
114
115	### `ts_features` (scalar)
116
117	Scalar over `(date, value)`. Same 116 outputs as struct fields.
118
119	### `ts_features_list` / `ts_features_table`
120
121	Discover the feature catalogue:
122
123	```sql
124	SELECT column_name, feature_name, parameter_suffix
125	FROM ts_features_list()
126	WHERE feature_name LIKE '%autocorr%';
127	```
128
129	### Custom feature subsets
130
131	Configure via JSON or CSV:
132
133	```sql
134	-- Load a JSON config
135	SELECT * FROM ts_features_by('sales', id, ds, y,
136	ts_features_config_from_json('{"features": ["mean", "std", "autocorrelation_lag1"]}'));
137
138	-- Or C

People who liked this also viewed

6 picks

Type

anofox-forecast-eda

35stars3forks

View on GitHub

How to install anofox-forecast-eda?

datazoode/anofox-forecast/anofox-forecast-eda

$npx -y skills add DataZooDE/anofox-forecast --skill anofox-forecast-eda

Installs into the current project.

Summary

Files · 1

View on GitHub

SKILL.md

1	# Anofox Forecast — EDA & Data Quality Cheat Sheet
2
3	Extension: `anofox_forecast` v0.15.3 \| DuckDB: v1.4.5 LTS / v1.5.4+ \| Dual naming: `ts_` and `anofox_fcst_ts_`
4
5	Understand your data before modelling it: per-series statistics, data-quality scores, feature vectors for ML.
6
7	## Statistics
8
9	### `ts_stats` (table function) — 34 metrics per series
10
11	```sql
12	ts_stats(source VARCHAR, group_col COLUMN, date_col COLUMN, value_col COLUMN,
13	frequency VARCHAR) → TABLE
14	```
15
16	Returns 34 columns: `length`, `n_nulls`, `n_nan`, `n_zeros`, `n_positive`, `n_negative`, `n_unique_values`, `is_constant`, `n_zeros_start`, `n_zeros_end`, `plateau_size`, `plateau_size_nonzero`, `mean`, `median`, `std_dev`, `variance`, `min`, `max`, `range`, `sum`, `skewness`, `kurtosis`, `tail_index`, `bimodality_coef`, `trimmed_mean`, `coef_variation`, `q1`, `q3`, `iqr`, `autocorr_lag1`, `trend_strength`, `seasonality_strength`, `entropy`, `stability`, plus date-derived `expected_length` and `n_gaps`.
17
18	```sql
19	-- Every stat for every series
20	SELECT * FROM ts_stats('sales', product_id, ds, y, '1d');
21
22	-- Quick sanity check
23	SELECT product_id, length, n_nulls, n_gaps, trend_strength, seasonality_strength
24	FROM ts_stats('sales', product_id, ds, y, '1d')
25	WHERE length < 30 OR n_gaps > 0
26	ORDER BY n_gaps DESC;
27	```
28
29	### `ts_stats_agg` (aggregate)
30
31	For custom `GROUP BY` shapes. Takes `(date, value)` — DO NOT wrap in `LIST(...)`.
32
33	```sql
34	ts_stats_agg(date_col TIMESTAMP, value_col DOUBLE)
35	→ STRUCT(length UBIGINT, n_nulls UBIGINT, …, mean DOUBLE, …)
36	```
37
38	```sql
39	SELECT product_id,
40	ts_stats_agg(ds, y).mean AS mean,
41	ts_stats_agg(ds, y).seasonality_strength AS seas_str
42	FROM sales GROUP BY product_id;
43	```
44
45	### `ts_stats_by` — alias for `ts_stats`
46
47	### `ts_stats_summary` — aggregate stats across all groups
48
49	Summarise the per-group stats into panel-level statistics (mean, std, min, max of each metric).
50
51	```sql
52	SELECT * FROM ts_stats_summary('sales', product_id, ds, y, '1d');
53	```
54
55	## Data quality
56
57	### `ts_data_quality` (table function) — per-series score card
58
59	```sql
60	ts_data_quality(source VARCHAR, unique_id_col COLUMN, date_col COLUMN, value_col COLUMN,
61	n_short INTEGER, frequency VARCHAR) → TABLE
62	```
63
64	Returns `unique_id` (group column renamed — the extension normalises it to `unique_id` in this output, even if the input col was `product_id`), plus `overall_score`, `structural_score`, `temporal_score`, `magnitude_score`, `behavioral_score`, `n_gaps`, `n_missing`, `is_constant`.
65
66	`n_short` is the series-length threshold below which a series is flagged short (typical: 14 for daily, 12 for monthly).
67
68	```sql
69	SELECT * FROM ts_data_quality('sales', product_id, ds, y, 14, '1d');
70
71	-- Rank series by quality (note: output col is `unique_id`, not `product_id`)
72	SELECT unique_id, overall_score, structural_score, temporal_score
73	FROM ts_data_quality('sales', product_id, ds, y, 14, '1d')
74	ORDER BY overall_score;
75	```
76
77	### `ts_data_quality_agg` (aggregate)
78
79	Same as `ts_stats_agg` — takes `(date, value)`, returns a STRUCT.
80
81	```sql
82	SELECT product_id,
83	ts_data_quality_agg(ds, y).overall_score AS q
84	FROM sales GROUP BY product_id;
85	```
86
87	### `ts_data_quality_summary` — panel-level roll-up
88
89	```sql
90	SELECT * FROM ts_data_quality_summary('sales', product_id, ds, y, 14);
91	```
92
93	### `ts_quality_report` — human-readable report
94
95	## Feature extraction (117 tsfresh-compatible features)
96
97	### `ts_features_by` — extract all 117 features
98
99	```sql
100	ts_features_by(source VARCHAR, group_col COLUMN, date_col COLUMN, value_col COLUMN) → TABLE
101	```
102
103	Returns the group column + 116 feature columns: `mean`, `standard_deviation`, `skewness`, `kurtosis`, `length`, `linear_trend_slope`, `autocorrelation_lag1`, …
104
105	```sql
106	-- Extract all features
107	SELECT * FROM ts_features_by('sales', product_id, ds, y);
108
109	-- Filter by feature values
110	SELECT product_id, mean, linear_trend_slope
111	FROM ts_features_by('sales', product_id, ds, y)
112	WHERE length > 30 AND autocorrelation_lag1 > 0.5;
113	```
114
115	### `ts_features` (scalar)
116
117	Scalar over `(date, value)`. Same 116 outputs as struct fields.
118
119	### `ts_features_list` / `ts_features_table`
120
121	Discover the feature catalogue:
122
123	```sql
124	SELECT column_name, feature_name, parameter_suffix
125	FROM ts_features_list()
126	WHERE feature_name LIKE '%autocorr%';
127	```
128
129	### Custom feature subsets
130
131	Configure via JSON or CSV:
132
133	```sql
134	-- Load a JSON config
135	SELECT * FROM ts_features_by('sales', id, ds, y,
136	ts_features_config_from_json('{"features": ["mean", "std", "autocorrelation_lag1"]}'));
137
138	-- Or C

People who liked this also viewed

6 picks

Type