End-to-end Data Science Platform
End-to-end Data Science Platform
Algolytics end-to-end data science platform helps to solve a variety of problems that financial intuitions face when trying to automate their business processes. The solution leverages AI & Machine Learning, geo scoring and behavioral biometrics capabilities to provide the complex credit and fraud score in less than 5 seconds.
The platform is distinguished by:
Comprehensiveness – covering all aspects of automating processes from data collection and processing, through Machine Learning model generation, to production launch and real-time monitoring
Flexibility and simplicity – fast deployment and modifications (no IT needed!) thanks to an intuitive user interface, drag&drop mechanisms and ready-to-use modules
High accuracy of risk assessment – ready-made risk models based on behavioral biometrics and location data with the market-verified quality (ROC - 84%, GINI 67%)
Ready integrations – ability to use multiple internal and external data sources, credit and business information bureaus, government registries and others
Platform applications
Data Quality & Location Intelligence - Automatically standardize, enrich and geocode addresses in the parcel / order deliveries or clean, standardize, deduplicate and enrich CRM / ERP customer data bases for optimizing sales & marketing activities.
Credit Scoring & Fraud Detection - Real-time scoring based on AI / Big Data, location risk scores and biometrics used in managed credit risk, BNPL, anti-fraud, early warning systems to optimize risk management.
Customer Intelligence - Customer segmentation, churn prevention Propensity models, Cross-selling / up-selling Recommendations, Lead generation / prioritization, In-house RTB/Programmatic marketing
Supply Chain Management & Manufacturing - Demand, sales planning & forecasting, Promotions planning & forecasting, Stock allocation and optimizations, Automatic stock replenishment, Route planning and optimization
Platform modules
Algolytics' comprehensive data science platform consists of five modules:
Data Quality - Data Quality & Location Intelligence
Event Engine - Feature store. Real time aggregation, processing and analysis of data streams
Advanced Miner - Data mining and modelling
Automatic Business Modeler (ABM) - Machine learning (AutoML)
Scoring.one - MLOps - scoring engine for recommendations and automated decision making
Modules integration
Event Engine (EE) -> ABM EE, in offline mode, utilizes the ABM REST API to automatically build models and generate scoring codes. Once saved in metadata, these codes are used for event scoring.
Event Engine (EE) -> Advanced Miner (AM) Models created with AM can be deployed in EE as scoring code. To do this, generate the scoring code for a selected model (choose Java as the code type) using the "ScoringCodeBuildTask" task. The generated scoring code can then be saved to the models table in the metadata.
Advanced Miner -> Scoring.one Models built with AM can be deployed in Scoring.one as scoring code. For a selected model, generate the scoring code (choosing Java as the code type) using the "ScoringCodeBuildTask" task. The generated scoring code can then be saved to a file and imported via the Scoring.one GUI (Scoring Code Management -> Scoring codes actions -> Upload new model). Scoring code deployment to Scoring.one can also be automated by preparing an appropriate script in AM that uses the Scoring.one REST API to perform the deployment operations.
Advanced Miner -> Scoring.one, Data Quality, Event Engine, ABM (REST API) In AM, you can integrate with the REST API by sending the necessary HTTP/HTTPS requests directly from the code.
ABM -> Scoring.one Models built with ABM can be deployed in Scoring.one as scoring code. The model export to Scoring.one can be done directly from the ABM GUI by clicking the "Deploy" button in the project view after building the model, or by downloading the scoring code (Java type) and importing it through the Scoring.one GUI (Scoring Code Management -> Scoring codes actions -> Upload new model).
Scoring.one -> Data Quality, Event Engine, ABM Scoring.one easily integrates with other services that expose their APIs on the internet. Using the dedicated "External data" node, you can easily make HTTP/HTTPS requests and integrate with the REST APIs of other Algolytics services (Event Engine, Data Quality, ABM).
Last updated
