iTuring ODA old

iTuring ODA: AI Powered Open Data Accelerator (Feature Store)

Ideate, Share, Discover Features

Open Data Accelerator for curating data pipelines

Financial services struggle to bring data from various sources, stich them together and use this data in the most beneficial way. Over 80% of the time in DSML lifecycle is spent on data preparation. Identifying, connecting multiple data sources and creating data pipelines, creating new features and curating data pipelines is a long process. Constructing new meaningful features are required to capture the essence of deep hidden behavior of the customer and select the right data lineage, which will ultimately optimize the inputs for a good model. Moreover, this does not remain a one-time activity as the created pipeline may not be reusable as it is, throwing up a big challenge, given the time and resources used, in creation and duplication of these features.

ML Data Pipeline for every business

With CyborgIntell iTuring’s ODA, connecting to multiple data sources becomes effortless with out-of-the-box connectors. Leverage 1000’s of pre-computed features to bolster your model accuracy and curate data pipelines for machine learning for your specific use case using drag and drop functions. Design your analytical solutions, select right modelling population and data lineage, allowing AI to create Mathematical, Statistical, Text Analytics, Pattern, Interaction and Value features. Reuse the pipeline for any use case in the domain and enable efficient re-training of models by only using the relevant feature pipeline. Users can compute feature pipelines instantly in real time, while eliminating duplicated efforts.

Ideate Discover Create your Data Science Projects

Data pipeline and preparation layer for machine learning

With ODA, you can ideate, share and discover features and create more effective machine learning pipelines. Perform data cleansing, aggregation, merging and efficient data preparation. Create auto-feature pipeline for ML development and deployment.

Innovative Analytical Design system

ODA helps you create observation and performance window and construct features with multiple data lineage. Functionality to create data lineage allows data scientist to experiment and improve efficacy of models. Faster feature computation, durability, and scalability. Create cohort, longitudinal and cross-sectional modelling population based on your use case. It is a system to design and create target/ input features and a configuration system to define time framework for data preparation.

AI driven Deep Features Synthesis

Open Data Accelerator enables creation of Mathematical, Statistical, Pattern, Text Analytics, Interaction, and value features. ODA offers pre-computed thousands of features to generate features from deep transaction data to understand customer’s transaction activity, habit and payment pattern, payment preference, utilization of channel, interaction with bank, depth and breadth of relationship, etc. It allows you to process credit bureau data and automatically compute thousands of features related to Customer Enquiry, Product, Revenue, Delinquency and Usage. Every feature computed from ODA can be traced, explained and validated. ODA also offers CLI based framework for data engineer to create their own feature recipe and plug it into feature pipeline.

Recipe to create new features

With over 10000+ preconfigured features ODA is one of the most impactful modules that decreases the time taken to create features, enabling faster data preparation. ODA also provides you with the recipe to create new online and offline features at any point of time. It just doesn’t stop there; you can also validate these features instantly.

Low latency serving API

Retrieving features for real time or batch predictions is a tedious task and hat can be prone to version and other errors. With ODA you can retrieve features with confidence, a high throughput batch API for creating point-in-time correct training data for both real time and batch predictions.

Connect to multiple sources of data

With ODA’s out-of-the-box connectors you can now connect to multiple sources of data instantly. Traditionally this activity takes a lot of time and effort, we have made it easy to connect with multiple sources and bring it together in one place making it extremely easy to create a data pipeline for analytics.