log_reduce_train_fl()

The function log_reduce_train_fl() finds common patterns in semi structured textual columns, such as log lines, and clusters the lines according to the extracted patterns. The function’s algorithm and most of the parameters are identical to log_reduce_fl(), but unlike log_reduce_fl() that outputs a patterns summary table, this function outputs the serialized model. The model can be used by the function log_reduce_predict_fl()/log_reduce_predict_full_fl() to predict the matched pattern for new log lines.

Syntax

T | invoke log_reduce_train_fl(reduce_col, model_name [, use_logram [, use_drain [, custom_regexes [, custom_regexes_policy [, delimiters [, similarity_th [, tree_depth [, trigram_th [, bigram_th ]]]]]]]]])

Parameters

The following parameters description is a summary. For more information, see More about the algorithm section.

Function definition

You can define the function by either embedding its code as a query-defined function, or creating it as a stored function in your database, as follows:

Query-defined

Define the function using the following let statement. No permissions are required.

let log_reduce_train_fl=(tbl:(*), reduce_col:string, model_name:string,
              use_logram:bool=True, use_drain:bool=True, custom_regexes: dynamic = dynamic([]), custom_regexes_policy: string = 'prepend',
              delimiters:dynamic = dynamic(' '), similarity_th:double=0.5, tree_depth:int = 4, trigram_th:int=10, bigram_th:int=15)
{
    let default_regex_table = pack_array('(/|)([0-9]+\\.){3}[0-9]+(:[0-9]+|)(:|)', '<IP>', 
                                         '([0-9a-fA-F]{8}-[0-9a-fA-F]{4}-[0-9a-fA-F]{4}-[0-9a-fA-F]{4}-[0-9a-fA-F]{12})', '<GUID>', 
                                         '(?<=[^A-Za-z0-9])(\\-?\\+?\\d+)(?=[^A-Za-z0-9])|[0-9]+$', '<NUM>');
    let kwargs = bag_pack('reduced_column', reduce_col, 'delimiters', delimiters,'output_column', 'LogReduce', 'parameters_column', '', 
                          'trigram_th', trigram_th, 'bigram_th', bigram_th, 'default_regexes', default_regex_table, 
                          'custom_regexes', custom_regexes, 'custom_regexes_policy', custom_regexes_policy, 'tree_depth', tree_depth, 'similarity_th', similarity_th, 
                          'use_drain', use_drain, 'use_logram', use_logram, 'save_regex_tuples_in_output', True, 'regex_tuples_column', 'RegexesColumn', 
                          'output_type', 'model');
    let code = ```if 1:
        from log_cluster import log_reduce
        result = log_reduce.log_reduce(df, kargs)
    ```;
    tbl
    | extend LogReduce=''
    | evaluate python(typeof(model:string), code, kwargs)
    | project name=model_name, timestamp=now(), model
};
// Write your query to use the function here.

Stored

Define the stored function once using the following .create function. Database User permissions are required.

.create-or-alter function with (folder = 'Packages\\Text', docstring = 'Find common patterns in textual logs, output a model')
log_reduce_train_fl(tbl:(*), reduce_col:string, model_name:string,
              use_logram:bool=True, use_drain:bool=True, custom_regexes: dynamic = dynamic([]), custom_regexes_policy: string = 'prepend',
              delimiters:dynamic = dynamic(' '), similarity_th:double=0.5, tree_depth:int = 4, trigram_th:int=10, bigram_th:int=15)
{
    let default_regex_table = pack_array('(/|)([0-9]+\\.){3}[0-9]+(:[0-9]+|)(:|)', '<IP>', 
                                         '([0-9a-fA-F]{8}-[0-9a-fA-F]{4}-[0-9a-fA-F]{4}-[0-9a-fA-F]{4}-[0-9a-fA-F]{12})', '<GUID>', 
                                         '(?<=[^A-Za-z0-9])(\\-?\\+?\\d+)(?=[^A-Za-z0-9])|[0-9]+$', '<NUM>');
    let kwargs = bag_pack('reduced_column', reduce_col, 'delimiters', delimiters,'output_column', 'LogReduce', 'parameters_column', '', 
                          'trigram_th', trigram_th, 'bigram_th', bigram_th, 'default_regexes', default_regex_table, 
                          'custom_regexes', custom_regexes, 'custom_regexes_policy', custom_regexes_policy, 'tree_depth', tree_depth, 'similarity_th', similarity_th, 
                          'use_drain', use_drain, 'use_logram', use_logram, 'save_regex_tuples_in_output', True, 'regex_tuples_column', 'RegexesColumn', 
                          'output_type', 'model');
    let code = ```if 1:
        from log_cluster import log_reduce
        result = log_reduce.log_reduce(df, kargs)
    ```;
    tbl
    | extend LogReduce=''
    | evaluate python(typeof(model:string), code, kwargs)
    | project name=model_name, timestamp=now(), model
}

Example

The following example uses the invoke operator to run the function.

Query-defined

To use a query-defined function, invoke it after the embedded function definition.

//
// Finding common patterns in HDFS logs, export and store the trained model in ML_Models table
//
.set-or-append ML_Models <|
//
let log_reduce_train_fl=(tbl:(*), reduce_col:string, model_name:string,
              use_logram:bool=True, use_drain:bool=True, custom_regexes: dynamic = dynamic([]), custom_regexes_policy: string = 'prepend',
              delimiters:dynamic = dynamic(' '), similarity_th:double=0.5, tree_depth:int = 4, trigram_th:int=10, bigram_th:int=15)
{
    let default_regex_table = pack_array('(/|)([0-9]+\\.){3}[0-9]+(:[0-9]+|)(:|)', '<IP>', 
                                         '([0-9a-fA-F]{8}-[0-9a-fA-F]{4}-[0-9a-fA-F]{4}-[0-9a-fA-F]{4}-[0-9a-fA-F]{12})', '<GUID>', 
                                         '(?<=[^A-Za-z0-9])(\\-?\\+?\\d+)(?=[^A-Za-z0-9])|[0-9]+$', '<NUM>');
    let kwargs = bag_pack('reduced_column', reduce_col, 'delimiters', delimiters,'output_column', 'LogReduce', 'parameters_column', '', 
                          'trigram_th', trigram_th, 'bigram_th', bigram_th, 'default_regexes', default_regex_table, 
                          'custom_regexes', custom_regexes, 'custom_regexes_policy', custom_regexes_policy, 'tree_depth', tree_depth, 'similarity_th', similarity_th, 
                          'use_drain', use_drain, 'use_logram', use_logram, 'save_regex_tuples_in_output', True, 'regex_tuples_column', 'RegexesColumn', 
                          'output_type', 'model');
    let code = ```if 1:
        from log_cluster import log_reduce
        result = log_reduce.log_reduce(df, kargs)
    ```;
    tbl
    | extend LogReduce=''
    | evaluate python(typeof(model:string), code, kwargs)
    | project name=model_name, timestamp=now(), model
};
HDFS_log_100k
| take 100000
| invoke log_reduce_train_fl(reduce_col="data", model_name="HDFS_100K")

Stored

//
// Finding common patterns in HDFS logs, export and store the trained model in ML_Models table
//
.set-or-append ML_Models <|
//
HDFS_log_100k
| take 100000
| invoke log_reduce_train_fl(reduce_col="data", model_name="HDFS_100K")

Output