dotData Enhances Data Science Automation Platform
March 12, 2019
released Version 1.4 of its dotData Data Science Automation
Platform. This latest update adds significant enhancements to the
platform and provides users with deeper insights, increased flexibility,
ease-of-use, and greater performance to meet their specific business
Key updates of the dotData
Platform Version 1.4 include:
• For example, for a retail store,
geo-temporal patterns, such as, "whether there is a sporting event
within three miles of the store during the next week," are often very
important to enable the store to optimize its inventory. dotData Version
1.4 enables users to automatically design such geo-temporal features
with a few clicks.
• dotData Version 1.4 now supports more state-of-the-art machine learning algorithms, including Gradient Boosting (XGBoost, LightGBM), Random Forest, and others. The Platform automatically tunes the hyperparameters of these algorithms to achieve the best performances in various statistical metrics.
• dotData users can automatically
take advantage of these highly-accurate ML algorithms, in addition to
previous white-box algorithms, to improve model accuracy.
• dotData Version 1.4 significantly enhances data preprocessing on both source data and features, including data integration, source data cleansing, and feature outlier filters, in addition to preprocessing functionalities supported in previous versions such as missing value imputation and data normalization.
• This data preprocessing is fully
automated, expanding the range of automation and further freeing up data
scientists to focus on the highest value projects with the biggest
• dotData Version 1.4 supports drag-and-drop data collection from CSV files in addition to existing JDBC data connectors. This enables users to import their locally-customized data quickly without handling SQL or interacting with databases.