Skip to content

Commit 71afeec

Browse files
szabosteveszabosteve
committed
Revert "[DOCS] Moves analysis resources to PUT DFA API docs (#50704)"
This reverts commit 4e1107d.
1 parent 4e1107d commit 71afeec

File tree

6 files changed

+315
-238
lines changed

6 files changed

+315
-238
lines changed

‎docs/reference/ml/df-analytics/apis/analysisobjects.asciidoc

Lines changed: 217 additions & 0 deletions
Original file line numberDiff line numberDiff line change
@@ -0,0 +1,217 @@
1+
[role="xpack"]
2+
[testenv="platinum"]
3+
[[ml-dfa-analysis-objects]]
4+
=== Analysis configuration objects
5+
6+
{dfanalytics-cap} resources contain `analysis` objects. For example, when you
7+
create a {dfanalytics-job}, you must define the type of analysis it performs.
8+
This page lists all the available parameters that you can use in the `analysis`
9+
object grouped by {dfanalytics} types.
10+
11+
12+
[discrete]
13+
[[oldetection-resources]]
14+
==== {oldetection-cap} configuration objects
15+
16+
An `outlier_detection` configuration object has the following properties:
17+
18+
`compute_feature_influence`::
19+
(Optional, boolean)
20+
include::{docdir}/ml/ml-shared.asciidoc[tag=compute-feature-influence]
21+
22+
`feature_influence_threshold`::
23+
(Optional, double)
24+
include::{docdir}/ml/ml-shared.asciidoc[tag=feature-influence-threshold]
25+
26+
`method`::
27+
(Optional, string)
28+
include::{docdir}/ml/ml-shared.asciidoc[tag=method]
29+
30+
`n_neighbors`::
31+
(Optional, integer)
32+
include::{docdir}/ml/ml-shared.asciidoc[tag=n-neighbors]
33+
34+
`outlier_fraction`::
35+
(Optional, double)
36+
include::{docdir}/ml/ml-shared.asciidoc[tag=outlier-fraction]
37+
38+
`standardization_enabled`::
39+
(Optional, boolean)
40+
include::{docdir}/ml/ml-shared.asciidoc[tag=standardization-enabled]
41+
42+
43+
[discrete]
44+
[[regression-resources]]
45+
==== {regression-cap} configuration objects
46+
47+
[source,console]
48+
--------------------------------------------------
49+
PUT _ml/data_frame/analytics/house_price_regression_analysis
50+
{
51+
"source": {
52+
"index": "houses_sold_last_10_yrs" <1>
53+
},
54+
"dest": {
55+
"index": "house_price_predictions" <2>
56+
},
57+
"analysis":
58+
{
59+
"regression": { <3>
60+
"dependent_variable": "price" <4>
61+
}
62+
}
63+
}
64+
--------------------------------------------------
65+
// TEST[skip:TBD]
66+
67+
<1> Training data is taken from source index `houses_sold_last_10_yrs`.
68+
<2> Analysis results will be output to destination index
69+
`house_price_predictions`.
70+
<3> The regression analysis configuration object.
71+
<4> Regression analysis will use field `price` to train on. As no other
72+
parameters have been specified it will train on 100% of eligible data, store its
73+
prediction in destination index field `price_prediction` and use in-built
74+
hyperparameter optimization to give minimum validation errors.
75+
76+
77+
[float]
78+
[[regression-resources-standard]]
79+
===== Standard parameters
80+
81+
`dependent_variable`::
82+
(Required, string)
83+
include::{docdir}/ml/ml-shared.asciidoc[tag=dependent-variable]
84+
+
85+
--
86+
The data type of the field must be numeric.
87+
--
88+
89+
`prediction_field_name`::
90+
(Optional, string)
91+
include::{docdir}/ml/ml-shared.asciidoc[tag=prediction-field-name]
92+
93+
`training_percent`::
94+
(Optional, integer)
95+
include::{docdir}/ml/ml-shared.asciidoc[tag=training-percent]
96+
97+
`randomize_seed`::
98+
(Optional, long)
99+
include::{docdir}/ml/ml-shared.asciidoc[tag=randomize-seed]
100+
101+
102+
[float]
103+
[[regression-resources-advanced]]
104+
===== Advanced parameters
105+
106+
Advanced parameters are for fine-tuning {reganalysis}. They are set
107+
automatically by <<ml-hyperparam-optimization,hyperparameter optimization>>
108+
to give minimum validation error. It is highly recommended to use the default
109+
values unless you fully understand the function of these parameters. If these
110+
parameters are not supplied, their values are automatically tuned to give
111+
minimum validation error.
112+
113+
`eta`::
114+
(Optional, double)
115+
include::{docdir}/ml/ml-shared.asciidoc[tag=eta]
116+
117+
`feature_bag_fraction`::
118+
(Optional, double)
119+
include::{docdir}/ml/ml-shared.asciidoc[tag=feature-bag-fraction]
120+
121+
`maximum_number_trees`::
122+
(Optional, integer)
123+
include::{docdir}/ml/ml-shared.asciidoc[tag=maximum-number-trees]
124+
125+
`gamma`::
126+
(Optional, double)
127+
include::{docdir}/ml/ml-shared.asciidoc[tag=gamma]
128+
129+
`lambda`::
130+
(Optional, double)
131+
include::{docdir}/ml/ml-shared.asciidoc[tag=lambda]
132+
133+
134+
[discrete]
135+
[[classification-resources]]
136+
==== {classification-cap} configuration objects
137+
138+
139+
[float]
140+
[[classification-resources-standard]]
141+
===== Standard parameters
142+
143+
`dependent_variable`::
144+
(Required, string)
145+
include::{docdir}/ml/ml-shared.asciidoc[tag=dependent-variable]
146+
+
147+
--
148+
The data type of the field must be numeric (`integer`, `short`, `long`, `byte`),
149+
categorical (`ip`, `keyword`, `text`), or boolean.
150+
--
151+
152+
`num_top_classes`::
153+
(Optional, integer)
154+
include::{docdir}/ml/ml-shared.asciidoc[tag=num-top-classes]
155+
156+
`prediction_field_name`::
157+
(Optional, string)
158+
include::{docdir}/ml/ml-shared.asciidoc[tag=prediction-field-name]
159+
160+
`training_percent`::
161+
(Optional, integer)
162+
include::{docdir}/ml/ml-shared.asciidoc[tag=training-percent]
163+
164+
`randomize_seed`::
165+
(Optional, long)
166+
include::{docdir}/ml/ml-shared.asciidoc[tag=randomize-seed]
167+
168+
169+
[float]
170+
[[classification-resources-advanced]]
171+
===== Advanced parameters
172+
173+
Advanced parameters are for fine-tuning {classanalysis}. They are set
174+
automatically by <<ml-hyperparam-optimization,hyperparameter optimization>>
175+
to give minimum validation error. It is highly recommended to use the default
176+
values unless you fully understand the function of these parameters. If these
177+
parameters are not supplied, their values are automatically tuned to give
178+
minimum validation error.
179+
180+
`eta`::
181+
(Optional, double)
182+
include::{docdir}/ml/ml-shared.asciidoc[tag=eta]
183+
184+
`feature_bag_fraction`::
185+
(Optional, double)
186+
include::{docdir}/ml/ml-shared.asciidoc[tag=feature-bag-fraction]
187+
188+
`maximum_number_trees`::
189+
(Optional, integer)
190+
include::{docdir}/ml/ml-shared.asciidoc[tag=maximum-number-trees]
191+
192+
`gamma`::
193+
(Optional, double)
194+
include::{docdir}/ml/ml-shared.asciidoc[tag=gamma]
195+
196+
`lambda`::
197+
(Optional, double)
198+
include::{docdir}/ml/ml-shared.asciidoc[tag=lambda]
199+
200+
[discrete]
201+
[[ml-hyperparam-optimization]]
202+
==== Hyperparameter optimization
203+
204+
If you don't supply {regression} or {classification} parameters, hyperparameter
205+
optimization will be performed by default to set a value for the undefined
206+
parameters. The starting point is calculated for data dependent parameters by
207+
examining the loss on the training data. Subject to the size constraint, this
208+
operation provides an upper bound on the improvement in validation loss.
209+
210+
A fixed number of rounds is used for optimization which depends on the number of
211+
parameters being optimized. The optimization starts with random search, then
212+
Bayesian optimization is performed that is targeting maximum expected
213+
improvement. If you override any parameters, then the optimization will
214+
calculate the value of the remaining parameters accordingly and use the value
215+
you provided for the overridden parameter. The number of rounds are reduced
216+
respectively. The validation error is estimated in each round by using 4-fold
217+
cross validation.

‎docs/reference/ml/df-analytics/apis/index.asciidoc

Lines changed: 2 additions & 0 deletions
Original file line numberDiff line numberDiff line change
@@ -14,6 +14,8 @@ You can use the following APIs to perform {ml} {dfanalytics} activities.
1414
* <<evaluate-dfanalytics,Evaluate {dfanalytics}>>
1515
* <<explain-dfanalytics,Explain {dfanalytics}>>
1616

17+
For the `analysis` object resources, check <<ml-dfa-analysis-objects>>.
18+
1719

1820
You can use the following APIs to perform {infer} operations.
1921

0 commit comments

Comments
 (0)