Community

Blog Events Webinars Tutorials Forum
  1. Blog
  2. Events
  3. Webinars
  4. Tutorials
  5. Forum
Create Account
×
Community Blog Data + AI Assists in Intelligent Analysis of Game Logs

Data + AI Assists in Intelligent Analysis of Game Logs

This article describes how to use the Data and AI solution of Alibaba Cloud to address challenges and stimulate the vitality of the gaming industry.

This article is the second article of the Get Started with Data + AI series published by Alibaba Cloud ApsaraDB. This series of articles describes the application scenarios of the Data + AI solution in various industries based on real customer cases and best practices.

>> Previous article: DMS + AnalyticDB Helps E-commerce Enterprises Make Decisions to Address Seven Major Challenges

This article describes how to use the Data and AI solution of Alibaba Cloud to address challenges and stimulate the vitality of the gaming industry.

1. Industry Trends

With the rapid development of the gaming industry, the amount of related data is increasing dramatically. The data increase is due not only to more players and longer online time but also to the various data-driven features integrated into games, such as personalized recommendation, dynamic difficulty adjustment, virtual economy system, real-time multi-player interaction, and behavior tracking and analysis. These features improve the gaming experience of players but also require the capability to process and store more data. This presents new challenges in data management and costs.

However, great values and opportunities also exist in massive data. In addition to supporting various data-driven features in games, gaming enterprises can also perform data analysis to better understand player behaviors and preferences. This helps enterprises optimize game design, improve player experience, and carry out precision marketing to increase revenues. With deeper data mining, enterprises can identify the key factors that affect player retention, find the best strategy to increase player engagement, and predict market trends and player behaviors, thereby enhancing the competitiveness and market influence of their products.

In this case, it becomes a main concern for enterprises to figure out how to focus on key data types, resolve management and technical issues, and explore data values in a continuous and efficient manner. This article describes how to use the Data and AI solution of Alibaba Cloud to address the challenges and stimulate the vitality of the gaming industry.

2. Technical Challenges

Of the massive data generated in the gaming industry, log data plays an important role. It records key information such as player behaviors, game status, and system performance. The information includes the logon and logoff time of players, growth trajectory of characters (such as character level-up), transaction details of virtual properties, interaction records between players (such as chats or team formation), and various events that occur in games (such as task completion). Log data also includes the technical data, such as the server response time and network latency. Log data can be widely used to analyze player experience, optimize product features, identify potential problems, and formulate operations strategies.

The amount of log data is surging and the analysis requirements are increasing. In this case, customers must find a cost-efficient solution to achieve a balance between performance and cost. Leading gaming enterprises may have the following requirements in the log analysis scenarios:

• Low data storage costs: Scenarios such as player complaint handling and compliance auditing require logs for backtracking. Therefore, full log data of games must be stored, which increases storage costs.

• High real-time query performance: To resolve unexpected issues at the earliest opportunity, massive data must be quickly processed and analyzed, which requires high analysis performance.

• Open source data: The business teams of the gaming enterprise use different analysis engines. In this case, the stored full log data must support access from multiple computing engines.

• Enhanced data insights based on integration with AI: Traditional business intelligence (BI) analysis focuses on the summary and display of historical data and helps decision makers understand the business performance within a period of time. However, it does not support performance prediction and customers must predict trends based on their experience. Based on the integration with AI, BI analysis can provide more accurate trend prediction on player behaviors, player payments, payment amounts, and player churn.

3. Data + AI Solution from Alibaba Cloud

This year, Alibaba Cloud ApsaraDB announced "DMS + X: Unified, Open, and Multimodal Data Management and Data Serving Driven by Data + AI". The platform simplifies data management and AI development and manages Data + AI development throughout its lifecycle. Based on Data Management (DMS) + X, Alibaba Cloud can help enterprises integrate data with AI, deploy business, and realize business values at the earliest opportunity.

To meet the preceding requirements in the gaming industry, Alibaba Cloud ApsaraDB provides DMS + AnalyticDB for MySQL as a solution.

_1

Real-time Access to Log Data

Cost-effective Game Log Archiving

Alibaba Cloud Simple Log Service (SLS) collects game logs from applications and provides the logs to downstream analysis engines for further processing and query. AnalyticDB for MySQL is a cloud-native data warehousing service developed by Alibaba Cloud ApsaraDB. It receives and stores real-time log data from SLS and ApsaraMQ for Kafka in warehouses or lakes and automatically generates metadata. Users can query data with ease on a GUI and select the destination storage based on their business requirements. When full game log data is written to lakes in the Iceberg format, the synchronization throughput can reach 1 GB per second and the data latency is less than 5 minutes. The storage cost of AnalyticDB for MySQL is as low as that of Object Storage Service (OSS), and lake management features such as data file merging, lifecycle management, and cache settings are available to improve the performance of lake data queries.

2

Efficient Log Data Query

AnalyticDB for MySQL provides two types of compute engines: the self-developed XIHE engine and the open source Spark engine. Customers can select different compute engines for data processing and analysis based on their business requirements. Lake tables support data analysis in seconds or minutes by using prefetch queries at the cache layer. Customers can also create warehouse tables to realize real-time analysis in sub-seconds.

3

Open Source Log Data

AnalyticDB for MySQL stores data in lakes in the Iceberg or Parquet format and provides API operations for Hive Metastore (HMS), OSS, and Hadoop Distributed File System (HDFS) engines to access data. A bucket that stores data can be attached to multiple instances for data sharing.

4

AI Implementation

In addition to BI analysis, DMS + AnalyticDB for MySQL can also process data characteristics, train models, evaluate and predict data, and apply AI models based on the log data stored in AnalyticDB for MySQL. AnalyticDB for MySQL supports end-to-end data processing and model development by using only SQL statements. Users can write SQL statements to import and train models and call remote inference services. AnalyticDB for MySQL also provides fully managed AI resource services that allow users to focus on business application development without the need to worry about the deployment of underlying resources.

The following sample code provides an example on how to create a model and perform prediction:

/* Create a model. */
create model bstdemo.bst 
options (
  model_type='xx',
  feature_cols=(event_list),
  target_cols=(target),
  hyperparameters = (
    use_best_ckpt = 'False',
    early_stopping_patience='0'
  )
)as select event_list, target from bstdemo.test;
/* Use regular functions to perform model-based prediction. */
SELECT ml_predict(
  '[db.]model_name:v2',  --  The name and version of the model.
   ['{options}'|NULL],  -- The additional settings, such as the maximum size of data in a batch. If no additional settings are required, enter NULL.
  -- The columns to be passed to the model. Enter a valid projection expression. The length of the expression is unlimited.
  <column_name1>,
  ...,
) as col_name FROM event_table;

DMS + AnalyticDB for MySQL supports integrated scheduling of data processing and model training, and also delivers automation capabilities for the entire model development lifecycle, including data access, data processing, model training, model publishing, and model-based inference. Core AI assets such as model performance data, model groups, and model versions are managed in the model center. Models can be deployed, rolled back, or published as online inference services, and associated with large language model (LLM) toolchains to develop AI applications. The DMS + AnalyticDB for MySQL solution supports fine-grained permission management on data assets and can be deployed in the virtual private clouds (VPCs) of customers, which ensures the security of data and models and protects customer privacy.

AI node resources and built-in algorithm models are provided for game log scenarios. Users can directly develop models in the DMS console without the need to build a machine learning platform. This reduces development costs.

For example, a leading gaming enterprise uses DMS + AnalyticDB for MySQL to predict the trends of player churn and player payments. The results show that the F1 score of the model is increased from 40% to a value greater than 70%.

5

4. Summary and Prospect

For game log storage, analysis, and prediction scenarios, Alibaba Cloud ApsaraDB provides a complete Data + AI solution and best practices of the solution to meet the requirements of enterprises for online data processing, real-time data analysis, AI implementation, and low development and O&M costs.

Data + AI provides a new way for enterprise growth. Enterprises must recognize the importance of Data + AI and use it as the key solution to promote intelligent transformation. This way, enterprises can remain competitive and leading in the market and better seize opportunities in the future.

The AI node of AnalyticDB for MySQL supports multiple development modes, such as MLSQL statements and Spark machine learning library (MLlib). With the existing analysis capabilities of AnalyticDB, AnalyticDB for MySQL provides a lightweight end-to-end solution for Data + AI applications.

Database AI Big Data Data analysis AnalyticDB Gaming Solution Gaming Industry
0 1 0
Share on

Read previous post:

DMS + AnalyticDB Helps E-commerce Enterprises Make Decisions to Address Seven Major Challenges

Read next post:

[Infographic] Highlights | Database New Features in January 2025

ApsaraDB

482 posts | 133 followers

You may also like

Comments

ApsaraDB

482 posts | 133 followers

Related Products

More Posts by ApsaraDB

See All