Microsoft Sequence Clustering Algorithm

Artikkeli
10/31/2023

Applies to: SQL Server 2019 and earlier Analysis Services Azure Analysis Services Fabric/Power BI Premium

Important

Data mining was deprecated in SQL Server 2017 Analysis Services and now discontinued in SQL Server 2022 Analysis Services. Documentation is not updated for deprecated and discontinued features. To learn more, see Analysis Services backward compatibility.

The Microsoft Sequence Clustering algorithm is a unique algorithm that combines sequence analysis with clustering. You can use this algorithm to explore data that contains events that can be linked in a sequence. The algorithm finds the most common sequences, and performs clustering to find sequences that are similar. The following examples illustrate the types of sequences that you might capture as data for machine learning, to provide insight about common problems or business scenarios:

Clickstreams or click paths generated when users navigate or browse a Web site
Logs that list events preceding an incident, such as a hard disk failure or server deadlock
Transaction records that describe the order in which a customer adds items to a online shopping cart
Records that follow customer or patient interactions over time, to predict service cancellations or other poor outcomes

This algorithm is similar in many ways to the Microsoft Clustering algorithm. However, instead of finding clusters of cases that contain similar attributes, the Microsoft Sequence Clustering algorithm finds clusters of cases that contain similar paths in a sequence.

Example

The Adventure Works Cycles web site collects information about what pages site users visit, and about the order in which the pages are visited. Because the company provides online ordering, customers must log in to the site. This provides the company with click information for each customer profile. By using the Microsoft Sequence Clustering algorithm on this data, the company can find groups, or clusters, of customers who have similar patterns or sequences of clicks. The company can then use these clusters to analyze how users move through the Web site, to identify which pages are most closely related to the sale of a particular product, and to predict which pages are most likely to be visited next.

How the Algorithm Works

The Microsoft Sequence Clustering algorithm is a hybrid algorithm that combines clustering techniques with Markov chain analysis to identify clusters and their sequences. One of the hallmarks of the Microsoft Sequence Clustering algorithm is that it uses sequence data. This data typically represents a series of events or transitions between states in a dataset, such as a series of product purchases or Web clicks for a particular user. The algorithm examines all transition probabilities and measures the differences, or distances, between all the possible sequences in the dataset to determine which sequences are the best to use as inputs for clustering. After the algorithm has created the list of candidate sequences, it uses the sequence information as an input for clustering using Expectation maximization (EM).

For a detailed description of the implementation, see Microsoft Sequence Clustering Algorithm Technical Reference.

Data Required for Sequence Clustering Models

When you prepare data for use in training a sequence clustering model, you should understand the requirements for the particular algorithm, including how much data is needed, and how the data is used.

The requirements for a sequence clustering model are as follows:

A single key column A sequence clustering model requires a key that identifies records.
A sequence column For sequence data, the model must have a nested table that contains a sequence ID column. The sequence ID can be any sortable data type. For example, you can use a Web page identifier, an integer, or a text string, as long as the column identifies the events in a sequence. Only one sequence identifier is allowed for each sequence, and only one type of sequence is allowed in each model.
Optional non sequence attributes The algorithm supports the addition of other attributes that are not related to sequencing. These attributes can include nested columns.

For example, in the example cited earlier of the Adventure Works Cycles Web site, a sequence clustering model might include order information as the case table, demographics about the specific customer for each order as non-sequence attributes, and a nested table containing the sequence in which the customer browsed the site or put items into a shopping cart as the sequence information.

For more detailed information about the content types and data types supported for sequence clustering models, see the Requirements section of Microsoft Sequence Clustering Algorithm Technical Reference.

Viewing a Sequence Clustering Model

The mining model that this algorithm creates contains descriptions of the most common sequences in the data. To explore the model, you can use the Microsoft Sequence Cluster Viewer. When you view a sequence clustering model, SQL Server Analysis Services shows you clusters that contain multiple transitions. You can also view pertinent statistics. For more information, see Browse a Model Using the Microsoft Sequence Cluster Viewer.

If you want to know more detail, you can browse the model in the Microsoft Generic Content Tree Viewer. The content stored for the model includes the distribution for all values in each node, the probability of each cluster, and details about the transitions. For more information, see Mining Model Content for Sequence Clustering Models (Analysis Services - Data Mining).

Creating Predictions

After the model has been trained, the results are stored as a set of patterns. You can use the descriptions of the most common sequences in the data to predict the next likely step of a new sequence. However, because the algorithm includes other columns, you can use the resulting model to identify relationships between sequenced data and inputs that are not sequential. For example, if you add demographic data to the model, you can make predictions for specific groups of customers. Prediction queries can be customized to return a variable number of predictions, or to return descriptive statistics.

For information about how to create queries against a data mining model, see Data Mining Queries. For examples of how to use queries with a sequence clustering model, see Sequence Clustering Model Query Examples.

Remarks

Does not support the use of Predictive Model Markup Language (PMML) to create mining models.
Supports drillthrough.
Supports the use of OLAP mining models and the creation of data mining dimensions.

Jaa