Decanter AI is an expert system that delivers fast, easy, accurate, and interpretable automated machine learning for enterprises. Decanter AI empowers data scientists and business analysts with world-class machine learning technologies through an intuitive visual interface, allowing enterprises to sole business challenges using an AI-driven approach by rapidly building highly accurate predictive models.
Traditionally, building machine learning models can easily take up to weeks or months from manually coding, testing, and evaluating just a few models. Decanter AI automatically builds and evaluates hundreds of models within hours, and models can be quickly retrained on-demand if necessary.
Using an automated data science workflow powered by an intuitive and code-free visual interface, Decanter AI makes it easy for anyone with basic machine learning knowledge to build predictive models comparable to those built by experienced data scientists.
Decanter AI builds and optimizes models from hundreds of data preprocessing, feature engineering techniques, machine learning algorithms and hyperparameter tuning to produce highly accurate predictive models.
Often, having an accurate model is often not enough for enterprise standards due to compliance and non-technical decision makers. Decanter AI provides extensive graphic visualizations and model explanations that can be easily understood in business terms.
Decanter AI uses a standardized workflow built on industry best practices to ensure consistent performance across users of different experience levels.
Exploratory Data Analysis
Decanter AI provides easy and effective ways to perform EDA to discover trends in datasets through statistical analysis, correlation heat map, distribution plots, correlation over time charts, and feature over time charts.
Auto Data Preprocessing
Decanter AI automatically performs feature transformation and standardization such as missing value imputation, count encoding, one-hot encoding, label encoding, and component encoding.
Auto Model Building
Decanter AI automatically builds and tunes models from over 60+ machine learning algorithms, and automatically performs stacked ensemble, balancing classes, and early stopping.
Auto Model Evaluation
Decanter AI automatically ranks models based on cross validation and holdout scores over different evaluators and provides ROC curves and correlation matrix for binary classification models.
Predictive models can be deployed into enterprise applications via REST API to make predictions. Models can also be used to predict datasets and prediction results can be visualized through hit rate charts.
Decanter AI ranks and displays the weighted importance of the features used in model training to help explain the model results.
Time Series Prediction
Decanter AI supports time series prediction that forecasts future values based on historical data and is crucial to many business problems involving time component.
Decanter AI uses a distributed deep learning framework that can spread its computing across different machines.
Decanter AI supports GPU acceleration, which can speed up the training time for models up to 50-100 times.
Decanter AI uses a Scala-based architecture to build algorithms that runs on Java Virtual Machine (JVM), which can perform up to 10-50 times faster than Python-based architecture and R-based architecture.
Decanter AI uses a scalable, distributed system that supports parallel processing, and can be deployed via one of three options: public cloud, on-premises server, and virtual machine (VM).