This adds machine learning model feature importance calculations to the inference processor.
The new flag in the configuration matches the analytics parameter name: `num_top_feature_importance_values`
Example:
```
"inference": {
"field_mappings": {},
"model_id": "my_model",
"inference_config": {
"regression": {
"num_top_feature_importance_values": 3
}
}
}
```
This will write to the document as follows:
```
"inference" : {
"feature_importance" : {
"FlightTimeMin" : -76.90955548511226,
"FlightDelayType" : 114.13514762158526,
"DistanceMiles" : 13.731580450792187
},
"predicted_value" : 108.33165831875137,
"model_id" : "my_model"
}
```
This is done through calculating the [SHAP values](https://arxiv.org/abs/1802.03888).
It requires that models have populated `number_samples` for each tree node. This is not available to models that were created before 7.7.
Additionally, if the inference config is requesting feature_importance, and not all nodes have been upgraded yet, it will not allow the pipeline to be created. This is to safe-guard in a mixed-version environment where only some ingest nodes have been upgraded.
NOTE: the algorithm is a Java port of the one laid out in ml-cpp: https://github.com/elastic/ml-cpp/blob/master/lib/maths/CTreeShapFeatureImportance.cc
usability blocked by: https://github.com/elastic/ml-cpp/pull/991
This commit modifies the codebase so that our production code uses a
single instance of the IndexNameExpressionResolver class. This change
is being made in preparation for allowing name expression resolution
to be augmented by a plugin.
In order to remove some instances of IndexNameExpressionResolver, the
single instance is added as a parameter of Plugin#createComponents and
PersistentTaskPlugin#getPersistentTasksExecutor.
Backport of #52596
Add enterprise operation mode to properly map enterprise license.
Aslo refactor XPackLicenstate class to consolidate license status and mode checks.
This class has many sychronised methods to check basically three things:
* Minimum operation mode required
* Whether security is enabled
* Whether current license needs to be active
Depends on the actual feature, either 1, 2 or all of above checks are performed.
These are now consolidated in to 3 helper methods (2 of them are new).
The synchronization is pushed down to the helper methods so actual checking
methods no longer need to worry about it.
resolves: #51081
When `PUT` is called to store a trained model, it is useful to return the newly create model config. But, it is NOT useful to return the inflated definition.
These definitions can be large and returning the inflated definition causes undo work on the server and client side.
Co-authored-by: Elastic Machine <elasticmachine@users.noreply.github.com>
This adds `_all` to Calendar searches. This enables users to supply the `_all` string in the `job_ids` array when creating a Calendar. That calendar will now be applied to all jobs (existing and newly created).
Closes#45013
Co-authored-by: Elastic Machine <elasticmachine@users.noreply.github.com>
This changes the tree validation code to ensure no node in the tree has a
feature index that is beyond the bounds of the feature_names array.
Specifically this handles the situation where the C++ emits a tree containing
a single node and an empty feature_names list. This is valid tree used to
centre the data in the ensemble but the validation code would reject this
as feature_names is empty. This meant a broken workflow as you cannot GET
the model and PUT it back
When changing a job state using a mechanism that doesn't
wait for the desired state to be reached within the production
code the test code needs to loop until the cluster state has
been updated.
Closes#52451
Following the change to store cluster state in Lucene indices
(#50907) it can take longer for all the cluster state updates
associated with node failure scenarios to be processed during
internal cluster tests where several nodes all run in the same
JVM.
ML mappings and index templates have so far been created
programmatically. While this had its merits due to static typing,
there is consensus it would be clear to maintain those in json files.
In addition, we are going to adding ILM policies to these indices
and the component for a plugin to register ILM policies is
`IndexTemplateRegistry`. It expects the templates to be in resource
json files.
For the above reasons this commit refactors ML mappings and index
templates into json resource files that are registered via
`MlIndexTemplateRegistry`.
Backport of #51765
This commit removes the need for DeprecatedRoute and ReplacedRoute to
have an instance of a DeprecationLogger. Instead the RestController now
has a DeprecationLogger that will be used for all deprecated and
replaced route messages.
Relates #51950
Backport of #52278
During a bug hunt, I caught a handful of things (unrelated to the bug) that could be potential issues:
1. Needlessly wrapping in exception handling (minor cleanup)
2. Potential of notifying listeners of a failure multiple times + even trying to notify of a success after a failure notification
In #51146 a rudimentary check for poor categorization was added to
7.6.
This change replaces that warning based on a Java-side check with
a new one based on the categorization_status field that the ML C++
sets. categorization_status was added in 7.7 and above by #51879,
so this new warning based on more advanced conditions will also be
in 7.7 and above.
Closes#50749
Changes the misleading error message when attempting to open
a job while the "cluster.persistent_tasks.allocation.enable"
setting is set to "none" to a clearer message that names the
setting.
Closes#51956
Refactors `DataFrameAnalyticsTask` to hold a `StatsHolder` object.
That just has a `ProgressTracker` for now but this is paving the
way to add additional stats like memory usage, analysis stats, etc.
Backport #52134
Employs `ResultsPersisterService` from `DataFrameRowsJoiner` in order
to add retries when a data frame analytics job is persisting the results
to the destination data frame.
Backport of #52048
This change adds support for the following new model_size_stats
fields:
- categorized_doc_count
- total_category_count
- frequent_category_count
- rare_category_count
- dead_category_count
- categorization_status
Backport of #51879
This commit changes how RestHandlers are registered with the
RestController so that a RestHandler no longer needs to register itself
with the RestController. Instead the RestHandler interface has new
methods which when called provide information about the routes
(method and path combinations) that are handled by the handler
including any deprecated and/or replaced combinations.
This change also makes the publication of RestHandlers safe since they
no longer publish a reference to themselves within their constructors.
Closes#51622
Co-authored-by: Jason Tedor <jason@tedor.me>
Backport of #51950
* [ML] Add bwc serialization unit test scaffold (#51889)
Adds new `AbstractBWCSerializationTestCase` which provides easy scaffolding for BWC serialization unit tests.
These are no replacement for true BWC tests (which execute actual old code). These tests do provide some good coverage for the current code when serializing to/from old versions.
* removing unnecessary override for 7.series branch
* adding necessary import
Co-authored-by: Elastic Machine <elasticmachine@users.noreply.github.com>
If the configs are removed (by some horrific means), we should still allow tasks to be cleaned up easily.
Datafeeds and jobs with missing configs are now visible in their respective _stats calls and can be stopped/closed.
Currently, the same class `FieldCapabilities` is used both to represent the
capabilities for one index, and also the merged capabilities across indices. To
help clarify the logic, this PR proposes to create a separate class
`IndexFieldCapabilities` for the capabilities in one index. The refactor will
also help when adding `source_path` information in #49264, since the merged
source path field will have a different structure from the field for a single index.
Individual changes:
* Add a new class IndexFieldCapabilities.
* Remove extra constructor from FieldCapabilities.
* Combine the add and merge methods in FieldCapabilities.Builder.
The work to switch file upload over to treating delimited files
like semi-structured text and using the ingest pipeline for CSV
parsing makes the multi-line start pattern used for delimited
files much more critical than it used to be.
Previously it was always based on the time field, even if that
was towards the end of the columns, and no multi-line pattern
was created if no timestamp was detected.
This change improves the multi-line start pattern by:
1. Never creating a multi-line pattern if the sample contained
only single line records. This improves the import
efficiency in a common case.
2. Choosing the leftmost field that has a well-defined pattern,
whether that be the time field or a boolean/numeric field.
This reduces the risk of a field with newlines occurring
earlier, and also means the algorithm doesn't automatically
fail for data without a timestamp.
This setting was introduced with the purpose of reducing the time took by
tests that shut nodes down. Tests like `MlDistributedFailureIT` and
`NetworkDisruptionIT`. However, it is unfortunate to have to set the value
to an explicit value in production. In addition, and most important, the dynamically
choosing the value for this setting makes it impossible to adopt static index template configs
that we register via `IndexTemplateRegistry`, which we need to use in order to start
registering ILM policies for the ML indices.
This commit removes this setting from our templates. I run the tests a few times and could
not see execution time differing significantly.
Backport of #51740
This adds logic to handle paging problems when the ID pattern + tags reference models stored as resources.
Most of the complexity comes from the issue where a model stored as a resource could be at the start, or the end of a page or when we are on the last page.
This commit switches the strategy for managing dot-prefixed indices that
should be hidden indices from using "fake" system indices to an explicit
exclusions list that must be updated when those indices are converted to
hidden indices.
* [ML][Inference] Fix weighted mode definition (#51648)
Weighted mode inaccurately assumed that the "max value" of the input values would be the maximum class value. This does not make sense.
Weighted Mode should know how many classes there are. Hence the new parameter `num_classes`. This indicates what the maximum class value to be expected.
Datafeeds being closed while starting could result in and NPE. This was
handled as any other failure, masking out the NPE. However, this
conflicts with the changes in #50886.
Related to #50886 and #51302
This commit deprecates the creation of dot-prefixed index names (e.g.
.watches) unless they are either 1) a hidden index, or 2) registered by
a plugin that extends SystemIndexPlugin. This is the first step
towards more thorough protections for system indices.
This commit also modifies several plugins which use dot-prefixed indices
to register indices they own as system indices, and adds a plugin to
register .tasks as a system index.
Changes the find_file_structure response to include a CSV
ingest processor in the ingest pipeline it suggests.
Previously the Kibana file upload functionality parsed CSV
in the browser, but by parsing CSV in the ingest pipeline
it makes the Kibana file upload functionality more easily
interchangable with Filebeat such that the configurations
it creates can more easily be used to import data with the
same structure repeatedly in production.
The DATE and DATESTAMP Grok patterns match 2 digit years
as well as 4 digit years. The pattern determination in
find_file_structure worked correctly in this case, but
the regex used to create a multi-line start pattern was
assuming a 4 digit year. Also, the quick rule-out
patterns did not always correctly consider 2 digit years,
meaning that detection was inconsistent.
This change fixes both problems, and also extends the
tests for DATE and DATESTAMP to check both 2 and 4 digit
years.
* [ML][Inference] add tags url param to GET (#51330)
Adds a new URL parameter, `tags` to the GET _ml/inference/<model_id> endpoint.
This parameter allows the list of models to be further reduced to those who contain all the provided tags.
As we prepare to introduce a new index for storing additional
information about data frame analytics jobs (e.g. intrumentation),
renaming this class to `DestinationIndex` better captures what it does
and leaves its prior name available for a more suitable use.
Backport of #51353
Co-authored-by: Elastic Machine <elasticmachine@users.noreply.github.com>
Data frame analytics classification currently only supports 2 classes for the
dependent variable. We were checking that the field's cardinality is not higher
than 2 but we should also check it is not less than that as otherwise the process
fails.
Backport of #51232
The ID of the datafeed's associated job was being obtained
frequently by looking up the datafeed task in a map that
was being modified in other threads. This could lead to
NPEs if the datafeed stopped running at an unexpected time.
This change reduces the number of places where a datafeed's
associated job ID is looked up to avoid the possibility of
failures when the datafeed's task is removed from the map
of running tasks during multi-step operations in other
threads.
Fixes#51285
There are two edge cases that can be ran into when example input is matched in a weird way.
1. Recursion depth could continue many many times, resulting in a HUGE runtime cost. I put a limit of 10 recursions (could be adjusted I suppose).
2. If there are no "fixed regex bits", exploring the grok space would result in a fence-post error during runtime (with assertions turned off)
Allows ML datafeeds to work with time fields that have
the "date_nanos" type _and make use of the extra precision_.
(Previously datafeeds only worked with time fields that were
exact multiples of milliseconds. So datafeeds would work
with "date_nanos" only if the extra precision over "date" was
not used.)
Relates #49889
This change introduces a new feature for indices so that they can be
hidden from wildcard expansion. The feature is referred to as hidden
indices. An index can be marked hidden through the use of an index
setting, `index.hidden`, at creation time. One primary use case for
this feature is to have a construct that fits indices that are created
by the stack that contain data used for display to the user and/or
intended for querying by the user. The desire to keep them hidden is
to avoid confusing users when searching all of the data they have
indexed and getting results returned from indices created by the
system.
Hidden indices have the following properties:
* API calls for all indices (empty indices array, _all, or *) will not
return hidden indices by default.
* Wildcard expansion will not return hidden indices by default unless
the wildcard pattern begins with a `.`. This behavior is similar to
shell expansion of wildcards.
* REST API calls can enable the expansion of wildcards to hidden
indices with the `expand_wildcards` parameter. To expand wildcards
to hidden indices, use the value `hidden` in conjunction with `open`
and/or `closed`.
* Creation of a hidden index will ignore global index templates. A
global index template is one with a match-all pattern.
* Index templates can make an index hidden, with the exception of a
global index template.
* Accessing a hidden index directly requires no additional parameters.
Backport of #50452
If 1000 different category definitions are created for a job in
the first 100 buckets it processes then an audit warning will now
be created. (This will cause a yellow warning triangle in the
ML UI's jobs list.)
Such a large number of categories suggests that the field that
categorization is working on is not well suited to the ML
categorization functionality.
Object fields cannot be used as features. At the moment _explain
API includes them and even worse it allows it does not error when
an object field is excluded. This creates the expectation to the
user that all children fields will also be excluded while it's not
the case.
This commit omits object fields from the _explain API and also
adds an error if an object field is included or excluded.
Backport of #51115
There have been occasional failures, presumably due to
too many tests running in parallel, caused by jobs taking
around 15 seconds to open. (You can see the job open
successfully during the cleanup phase shortly after the
failure of the test in these cases.) This change increases
the wait time from 10 seconds to 20 seconds to reduce the
risk of this happening.
Check it out:
```
$ curl -u elastic:password -HContent-Type:application/json -XPOST localhost:9200/test/_update/foo?pretty -d'{
"dac": {}
}'
{
"error" : {
"root_cause" : [
{
"type" : "x_content_parse_exception",
"reason" : "[2:3] [UpdateRequest] unknown field [dac] did you mean [doc]?"
}
],
"type" : "x_content_parse_exception",
"reason" : "[2:3] [UpdateRequest] unknown field [dac] did you mean [doc]?"
},
"status" : 400
}
```
The tricky thing about implementing this is that x-content doesn't
depend on Lucene. So this works by creating an extension point for the
error message using SPI. Elasticsearch's server module provides the
"spell checking" implementation.
s
* [ML][Inference] Adding classification_weights to ensemble models
classification_weights are a way to allow models to
prefer specific classification results over others
this might be advantageous if classification value
probabilities are a known quantity and can improve
model error rates.
In classes where the client is used directly rather than through a call to
executeAsyncWithOrigin explicitly require the client to be OriginSettingClient
rather than using the Client interface.
Also remove calls to deprecated ClientHelper.clientWithOrigin() method.
Adds a new parameter to regression and classification that enables computation
of importance for the top most important features. The computation of the importance
is based on SHAP (SHapley Additive exPlanations) method.
Backport of #50914
The system created and models we provide now use the `_xpack` user for uniformity with our other features
The `PUT` action is now an admin cluster action
And XPackClient class now references the action instance.
* [ML][Inference] PUT API (#50852)
This adds the `PUT` API for creating trained models that support our format.
This includes
* HLRC change for the API
* API creation
* Validations of model format and call
* fixing backport
This commit removes validation logic of source and dest indices
for data frame analytics and replaces it with using the common
`SourceDestValidator` class which is already used by transforms.
This way the validations and their messages become consistent
while we reduce code.
This means that where these validations fail the error messages
will be slightly different for data frame analytics.
Backport of #50841
A very large number of recursive calls can cause a stack overflow
exception. This commit forks the recursive calls for non-async
processors. Once forked, each thread will handle at most 10
recursive calls to help keep the stack size and thread count
down to a reasonable size.
This adds the necessary named XContent classes to the HLRC for the lang ident model. This is so the HLRC can call `GET _ml/inference/lang_ident_model_1?include_definition=true` without XContent parsing errors.
The constructors are package private as since this classes are used exclusively within the pre-packaged model (and require the specific weights, etc. to be of any use).
This PR adds per-field metadata that can be set in the mappings and is later
returned by the field capabilities API. This metadata is completely opaque to
Elasticsearch but may be used by tools that index data in Elasticsearch to
communicate metadata about fields with tools that then search this data. A
typical example that has been requested in the past is the ability to attach
a unit to a numeric field.
In order to not bloat the cluster state, Elasticsearch requires that this
metadata be small:
- keys can't be longer than 20 chars,
- values can only be numbers or strings of no more than 50 chars - no inner
arrays or objects,
- the metadata can't have more than 5 keys in total.
Given that metadata is opaque to Elasticsearch, field capabilities don't try to
do anything smart when merging metadata about multiple indices, the union of
all field metadatas is returned.
Here is how the meta might look like in mappings:
```json
{
"properties": {
"latency": {
"type": "long",
"meta": {
"unit": "ms"
}
}
}
}
```
And then in the field capabilities response:
```json
{
"latency": {
"long": {
"searchable": true,
"aggreggatable": true,
"meta": {
"unit": [ "ms" ]
}
}
}
}
```
When there are no conflicts, values are arrays of size 1, but when there are
conflicts, Elasticsearch includes all unique values in this array, without
giving ways to know which index has which metadata value:
```json
{
"latency": {
"long": {
"searchable": true,
"aggreggatable": true,
"meta": {
"unit": [ "ms", "ns" ]
}
}
}
}
```
Closes#33267
Switch from a 32 bit Java hash to a 128 bit Murmur hash for
creating document IDs from by/over/partition field values.
The 32 bit Java hash was not sufficiently unique, and could
produce identical numbers for relatively common combinations
of by/partition field values such as L018/128 and L017/228.
Fixes#50613
The end offset of a tokenizer is supposed to point one past the
end of the input, not to the end character of the input. The
ml_classic tokenizer was erroneously doing the latter.
Sharing a random generator may cause test failures as non-threadsafe random generators are periodically utilized in tests (see: https://github.com/elastic/elasticsearch/issues/50651)
This change constructs a calls `Randomness.get()` within the `bulkIndexWithRetry` method so that the returned `Random` object is only used in a single thread. Before, the member variable could have been used between threads, which caused test failures.
* [ML][Inference] lang_ident model (#50292)
This PR contains a java port of Google's CLD3 compact NN model https://github.com/google/cld3
The ported model is formatted to fit within our inference model formatting and stored as a resource in the `:xpack:ml:` plugin and is under basic license.
The model is broken up into two major parts:
- Preprocessing through the custom embedding (based on CLD3's embedding layer)
- Pushing the embedded text through the two layers of fully connected shallow NN.
Main differences between this port and CLD3:
- We take advantage of Java's internal Unicode handling where possible (i.e. codepoints, characters, decoders, etc.)
- We do not trim down input text by removing duplicated tokens
- We do not encode doubles/floats as longs/integers.
Adds a `force` parameter to the delete data frame analytics
request. When `force` is `true`, the action force-stops the
jobs and then proceeds to the deletion. This can be used in
order to delete a non-stopped job with a single request.
Closes#48124
Backport of #50553
Eclipse 4.13 shows a type mismatch error in the affected line because it cannot
correctly infer the boolean return type for the method call. Assigning return
value to a local variable resolves this problem.
In order to ensure any persisted model state is searchable by the moment
the job reports itself as `stopped`, we need to refresh the state index
before completing.
This should fix the occasional failures we see in #50168 and #50313 where
the model state appears missing.
Closes#50168Closes#50313
Backport of #50322
This fixes support for nested fields
We now support fully nested, fully collapsed, or a mix of both on inference docs.
ES mappings allow the `_source` to be any combination of nested objects + dot delimited fields.
So, we should do our best to find the best path down the Map for the desired field.
This commit adds removal of unused data frame analytics state
from the _delete_expired_data API (and in extend th ML daily
maintenance task). At the moment the potential state docs
include the progress document and state for regression and
classification analyses.
Backport of #50243
Executing the data frame analytics _explain API with a config that contains
a field that is not in the includes list but at the same time is the excludes
list results to trying to remove the field twice from the iterator. That causes
an `IllegalStateException`. This commit fixes this issue and adds a test that
captures the scenario.
Backport of #50192
This adds a new field for the inference processor.
`warning_field` is a place for us to write warnings provided from the inference call. When there are warnings we are not going to write an inference result. The goal of this is to indicate that the data provided was too poor or too different for the model to make an accurate prediction.
The user could optionally include the `warning_field`. When it is not provided, it is assumed no warnings were desired to be written.
The first of these warnings is when ALL of the input fields are missing. If none of the trained fields are present, we don't bother inferencing against the model and instead provide a warning stating that the fields were missing.
Also, this adds checks to not allow duplicated fields during processor creation.
This exchanges the direct use of the `Client` for `ResultsPersisterService`. State doc persistence will now retry. Failures to persist state will still not throw, but will be audited and logged.
This commit fixes a bug that caused the data frame analytics
_explain API to time out in a multi-node setup when the source
index was missing. When we try to create the extracted fields detector,
we check the index settings. If the index is missing that responds
with a failure that could be wrapped as a remote exception.
While we unwrapped correctly to check if the cause was an
`IndexNotFoundException`, we then proceeded to cast the original
exception instead of the cause.
Backport of #50176
* [ML] Add graceful retry for anomaly detector result indexing failures (#49508)
All results indexing now retry the amount of times configured in `xpack.ml.persist_results_max_retries`. The retries are done in a semi-random, exponential backoff.
* fixing test
Benjamin Trent
Jack Conradson
Jay Modi
Przemysław Witek
Yang Wang
David Kyle
David Roberts
Dimitris Athanasiou
Julie Tibshirani
Gordon Brown
Henning Andersen
Tom Veasey
Nik Everett
Jake Landis
Adrien Grand
Christoph Büscher