Skip to content

Commit 6de0d84

Browse files
ahoslerahosler
committed
add yaml schema table
1 parent 030dadc commit 6de0d84

File tree

1 file changed

+215
-0
lines changed

1 file changed

+215
-0
lines changed

‎docs/source/user_guide/operators/forecast_operator/yaml_schema.rst

Lines changed: 215 additions & 0 deletions
Original file line numberDiff line numberDiff line change
@@ -19,6 +19,221 @@ Below is an example of a ``forecast.yaml`` file with every parameter specified:
1919
horizon: 3
2020
target_column: target
2121
22+
23+
24+
.. list-table:: Forecast Operator Configuration Reference
25+
:widths: 20 10 10 20 40
26+
:header-rows: 1
27+
28+
* - Field
29+
- Type
30+
- Required
31+
- Default
32+
- Description
33+
34+
* - historical_data
35+
- dict
36+
- Yes
37+
- {"url": "data.csv"}
38+
- Indexed by date and optionally target category. Includes targets and endogeneous data.
39+
40+
* - additional_data
41+
- dict
42+
- No
43+
-
44+
- Optional exogeneous data. Must align with historical_data structure.
45+
46+
* - test_data
47+
- dict
48+
- No
49+
-
50+
- Optional, used for evaluation if provided.
51+
52+
* - output_directory
53+
- dict
54+
- No
55+
-
56+
- Where output files will be saved. Accepts the same data schema as inputs.
57+
58+
* - report_filename
59+
- string
60+
- No
61+
- report.html
62+
- Output report file name.
63+
64+
* - report_title
65+
- string
66+
- No
67+
-
68+
- Title of the output report.
69+
70+
* - report_theme
71+
- string
72+
- No
73+
- light
74+
- Theme of the report. Options: light, dark.
75+
76+
* - metrics_filename
77+
- string
78+
- No
79+
- metrics.csv
80+
- Filename for metrics output.
81+
82+
* - test_metrics_filename
83+
- string
84+
- No
85+
- test_metrics.csv
86+
- Filename for test set evaluation metrics.
87+
88+
* - forecast_filename
89+
- string
90+
- No
91+
- forecast.csv
92+
- Output forecast data file.
93+
94+
* - global_explanation_filename
95+
- string
96+
- No
97+
- global_explanations.csv
98+
- File for global explanations.
99+
100+
* - local_explanation_filename
101+
- string
102+
- No
103+
- local_explanations.csv
104+
- File for local explanations.
105+
106+
* - target_column
107+
- string
108+
- Yes
109+
- target
110+
- Column to forecast.
111+
112+
* - datetime_column.name
113+
- string
114+
- Yes
115+
- Date
116+
- Timestamp column name.
117+
118+
* - datetime_column.format
119+
- string
120+
- No
121+
-
122+
- Optional datetime format.
123+
124+
* - target_category_columns
125+
- list
126+
- No
127+
- ["Series ID"]
128+
- Categories for multi-series forecasting.
129+
130+
* - horizon
131+
- integer
132+
- Yes
133+
- 1
134+
- Forecast horizon (how far ahead).
135+
136+
* - model
137+
- string
138+
- No
139+
- prophet
140+
- Model to use. Options: prophet, arima, neuralprophet, automlx, autots, auto-select.
141+
142+
* - model_kwargs
143+
- dict
144+
- No
145+
-
146+
- Parameters specific to the chosen model.
147+
148+
* - preprocessing.enabled
149+
- boolean
150+
- No
151+
- true
152+
- Whether to apply preprocessing.
153+
154+
* - preprocessing.steps.missing_value_imputation
155+
- boolean
156+
- No
157+
- true
158+
- Impute missing values.
159+
160+
* - preprocessing.steps.outlier_treatment
161+
- boolean
162+
- No
163+
- false
164+
- Handle outliers.
165+
166+
* - generate_explanations
167+
- boolean
168+
- No
169+
- false
170+
- Toggle local and global explanations.
171+
172+
* - explanations_accuracy_mode
173+
- string
174+
- No
175+
- FAST_APPROXIMATE
176+
- Explanation mode. Options: HIGH_ACCURACY, BALANCED, FAST_APPROXIMATE, AUTOMLX.
177+
178+
* - generate_report
179+
- boolean
180+
- No
181+
- true
182+
- Enable report generation.
183+
184+
* - generate_metrics
185+
- boolean
186+
- No
187+
- true
188+
- Enable metrics file generation.
189+
190+
* - metric
191+
- string
192+
- No
193+
- MAPE
194+
- Evaluation metric. Options: MAPE, RMSE, MSE, SMAPE (case-insensitive).
195+
196+
* - what_if_analysis
197+
- dict
198+
- No
199+
-
200+
- Save models to model catalog if enabled. Includes deployment config.
201+
202+
* - previous_output_dir
203+
- string
204+
- No
205+
-
206+
- Load previous run outputs.
207+
208+
* - generate_model_parameters
209+
- boolean
210+
- No
211+
-
212+
- Export fitted model parameters.
213+
214+
* - generate_model_pickle
215+
- boolean
216+
- No
217+
-
218+
- Export trained model as pickle file.
219+
220+
* - confidence_interval_width
221+
- float
222+
- No
223+
- 0.80
224+
- Width of confidence intervals in forecast.
225+
226+
* - tuning.n_trials
227+
- integer
228+
- No
229+
- 10
230+
- Number of tuning trials for hyperparameter search.
231+
232+
233+
Further Description
234+
-------------------
235+
236+
22237
* **kind**: The YAML file always starts with ``kind: operator``. This identifies the type of service. Common kinds include ``operator`` and ``job``, but here, ``operator`` is required.
23238
* **type**: The type of operator is ``forecast``, which should always be specified when using this forecast operator.
24239
* **version**: The only available version is ``v1``.

0 commit comments

Comments
 (0)