[
https://issues.apache.org/jira/browse/MADLIB-1220?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16412159#comment-16412159
]
Jingyi Mei commented on MADLIB-1220:
------------------------------------
For grouping, we proposed the following implementation:
# when packing rows from source table, we apply partition by grouping_cols for
row_number() as row_id and group by row_id while doing matrix_agg, so that we
pack the rows separately for different groups with each group has its own
row_ids. See this query for more info.
https://github.com/apache/madlib/blob/master/src/ports/postgres/modules/utilities/minibatch_preprocessing.py_in#L128
# apply group by for standardization. In class MiniBatchStandardizer, instead
of getting x_mean_str and x_std_dev_str as an array string for the whole
dataset, we call a function which saves mean and standard deviation arrays by
group in a temp table, and then we call madlib.utils_normalize_data to
normalize data by joining the temp table and source table on grouping column.
# Because the temp table mentioned in step 2 contains all the info we need for
standardization output table, we decided to make it a permanent table instead
of temp, so that when calling grouping we don't need to create standardization
output table again by doing another table scan. See this line for more info.
https://github.com/apache/madlib/blob/master/src/ports/postgres/modules/convex/utils_regularization.py_in#L107
> Pre-processing helper function for mini-batching - grouping
> ------------------------------------------------------------
>
> Key: MADLIB-1220
> URL: https://issues.apache.org/jira/browse/MADLIB-1220
> Project: Apache MADlib
> Issue Type: New Feature
> Components: Module: Utilities
> Reporter: Nikhil
> Assignee: Nikhil
> Priority: Major
> Fix For: v1.14
>
>
> Related to
> https://issues.apache.org/jira/browse/MADLIB-1200
> Story
> {{As a}}
> data scientist
> {{I want to}}
> add grouping to mini-batch pre-process
> {{so that}}
> I can handle groups with a single operation.
> Interface
> {code:java}
> minibatch_preprocessor(
> source_table, -- Name of the table containing input data
> output_table, -- Name of the output table for mini-batching
> dependent_varname, -- Name of the dependent variable column
> independent_varname, -- Expression list to evaluate for the independent
> variables
> grouping_cols, -- Preprocess separately by group
> buffer_size -- Number of source input rows to pack into batch
> )
> {code}
> where
> {code:java}
> source_table
> TEXT. Name of the table containing input data. Can also be a view.
> output_table
> TEXT. Name of the output table from the preprocessor which will be used as
> input to algorithms that support mini-batching.
> dependent_varname
> TEXT. Column name or expression to evaluate for the dependent variable.
> independent_varname
> TEXT. Column name or expression list to evaluate for the independent
> variable. Will be cast to double when packing.
> grouping_cols (optional)
> TEXT, default: NULL. An expression list used to group the input dataset into
> discrete groups, running one preprocessing step per group. Similar to the SQL
> GROUP BY clause. When this value is NULL, no grouping is used and a single
> preprocessing step is performed for the whole data set.
> buffer_size (optional) INTEGER, default: ???. Number of source input rows to
> pack into batch.
> {code}
> The output table contains the following columns:
> {code:java}
> id INTEGER. Unique id for packed table.
> dependent_varname FLOAT8[]. Packed array of dependent
> variables.
> independent_varname FLOAT8[]. Packed array of independent
> variables.
> grouping_cols TEXT. Name of grouping columns.
> {code}
> A summary table named <output_table>_summary is created together with the
> output table. It has the following columns:
> {code:java}
> source_table Source table name.
> output_table Output table name from preprocessor.
> dependent_varname Dependent variable.
> independent_varname Independent variables.
> buffer_size Buffer size used in preprocessing step.
> dependent_vartype “Continuous” or “Categorical”
> class_values Class values of the dependent variable (NULL
> for continuous vars).
> num_rows_processed The total number of rows that were used in the
> computation.
> num_missing_rows_skipped The total number of rows that were skipped
> because of NULL values in them.
> grouping_cols Names of the grouping columns.
> {code}
> A standardization table named <output_table>_standardization is created
> together with the output table. It has the following columns:
> {code:java}
> <grouping_col_expression> Group
> mean Mean of independent vars by group
> std Standard deviation of independent vars
> by group
> {code}
>
> Acceptance
--
This message was sent by Atlassian JIRA
(v7.6.3#76005)
