AdvancedMiner

AdvancedMiner is a data science platform designed to unify the full range of analytical tasks, including data processing, modeling, validation, and analysis, within a single integrated environment. The platform is aimed at data science and machine learning teams seeking to streamline the creation, deployment, and monitoring of AI and machine learning processes. AdvancedMiner supports every stage of the data mining process through a graphical workspace, facilitating collaboration and efficient workflow management.

The platform offers a comprehensive suite of data processing capabilities, such as extraction from various database systems and files, handling large datasets with thousands of variables and millions of records, and providing built-in systems for data quality improvement, including standardization, validation, and enrichment. Users can merge datasets, sample data, split data into training, validation, and test sets, and perform variable selection and transformation using a variety of algorithms and techniques like normalization, principal component analysis, and imputation. Interactive previews of variable distributions and data export with editable charts to MS Excel are also supported. For Polish-language data, features include text data indexing, tokenization, lemmatization, sentiment analysis, synonym dictionaries, and named entity extraction.

AdvancedMiner's modeling capabilities encompass a wide array of algorithms for classification (such as logistic regression, decision trees, random forests, XGBoost, neural networks), regression, segmentation (including k-means and Kohonen SOM), time series analysis, survival analysis, and graph/network analysis. The platform enables the creation of custom statistical algorithms, interactive model visualization, and integration of models with external IT applications via scoring code. Models can be exported as scoring code for deployment, and the platform supports both batch and real-time predictions. Validation features include versioning, experiment tracking, variable significance analysis, discriminant analysis, a range of model quality metrics (ACC, AUC, MSE), visual model comparison tools (LIFT, ROC), and numerous statistical tests.

The infrastructure is built on a client-server architecture supporting parallel computation, automatic queuing, and team collaboration. AdvancedMiner includes a proprietary high-performance database (GDBase) with no table size limitations and an extensive library of analytical functions. The platform allows connection to external databases via JDBC/ODBC, enabling data import from various technologies and integration across multiple servers and clients. It is particularly suited for applications such as credit scoring, fraud detection, location intelligence, vendor verification, and customer intelligence, especially where precise address and spatial data are required.