dotDATA Overview: Enterprise AI Platform

Dominick Amalraj - Consultant

Machine learning objectives are transforming the way organizations are conducting business. With the ability to understand what the future holds more clearly and insights that can change the way an enterprise operates, it is no shock there is a desire to implement these solutions. Machine learning use cases are difficult to deliver on due to the skillset needed and the time-consuming nature associated with these projects. Automated machine learning has opened new doors for both data scientists and non-data scientists to more efficiently take on these projects. Once a dataset is prepared, automated machine learning can implement top-notch models at the fraction of the time and allow users to deploy them through REST API or other means.

But even though auto-ml platforms are saving time and efficiency in the model building phase, there is still a large amount of work that goes into the various other phases in the data science process. The 80/20 rule in data science says 80% of the time spent on a data science project is on finding, cleaning, and organizing data. This leaves only 20% of the time to actually build out predictive models. Within the data preparation stage, data science teams conduct various ETL steps to create their data table, clean the data, and conduct feature engineering. dotData, an Enterprise AI platform, has created a solution that not only automates model building but the entire data science process.

The dotData platform conducts and automates all the necessary steps in the machine learning process as the image above depicts. The steps automated within this platform include the ETL process, feature engineering, model building, and production of final results. This article will cover the main functionalities of the dotData platform.

1. Automated ETL Process

In any machine learning project, the ETL process is a vital piece. This includes bringing all the necessary tables from various data sources, cleaning the data, and creating a table which can be used for model training. dotData allows users to upload data tables into the platform from a plethora of sources. Once the data is ingested, dotData can automatically clean, analysis, and prepare multi-dimensional tables ready for model building.

The user is in charge of selecting a feature to predict such as sales or churn. From there, the dotData platform will be able to join the necessary tables needed based on the statistical relationship to the target field and conduct the necessary ETL steps to form a training dataset. This removes the need for a user to evaluate which tables are necessary for model building and any ETL steps needed to configure the final table. On top of that, dotData will automatically cleanse the schema and final data table used for model building.

2. AI – Powered Feature Engineering

Another crucial aspect in machine learning is the feature engineering. There is quite a bit that can be done in this step including: computing new features, conducting any necessary transformations, removing redundant or insignificant features, and more. This can be a very complex and lengthy process, but with dotData this would only take a few hours. dotData presents the features and most relevant findings with thorough visual explanations. This gives users a better understanding of the main drivers that impact a target field.

3. Auto-ML and Production Capabilities

dotData has plenty of innovative machine learning techniques when it comes to their automated machine learning. This includes models from XGBoost, LightGMB, TensorFlow, and PyTorch. As with many automated machine learning tools, the platform will automatically tune and reconfigure models to achieve the most accurate results. From the thorough feature engineering conducted in the dotData platform, the model’s accuracy sees significant improvement and insights are more clearly illustrated. After models are made, dotData has a GUI that allows users to see insights from the models, make predictions, and gain the necessary information to make better business decisions. With its REST API, predictions made in dotData can be accessible to a number of BI platforms and sources.

dotData is focused on limiting the effort and time associated in the machine learning process. With the reduced time to value from automating the entire data science process, users will be able to discover more from their models and dive deeper into them to better prepare for the future. Additionally, with the reduced time per project, data science teams are able to tackle more use cases than ever before. If you have any questions or would like to schedule a demo of dotData please reach out to for more information