As companies scale their use of advanced analytics and AI, more people are involved in the design, development, and consumption of AI. Dataiku 9 has exciting new capabilities to help business people contribute their knowledge to AI projects and build trust in AI outputs. Application developers also get a boost with the addition of the Dash framework.
Data Preparation Gets Even Smarter
For business analysts, data preparation can be a time-consuming task. Anything that streamlines the process and creates higher quality data drives value for business teams. Dataiku 9 streamlines data preparation in the following areas:
- Fuzzy Join - Join datasets based on geospatial distance, Euclidean distance, or text similarity
- Regex Builder (Smart Pattern Builder) - Create regular expressions by highlighting text in fields and then automatically extract that text into new columns
- Date Handling - Easier date parsing to extract information from date columns
- Formula Editor - Contextual help and examples when developing formulas
Want to see the data prep improvements in action? Check out the video at the end of this post for a demonstration and discussion on these capabilities.
Smart pattern builder for regex in Dataiku 9
More AI Application Options for Developers
AI can drive enhanced customer experiences and better business decisions, but only if users have access to AI in a context that makes sense to them. Dataiku 8 included integrations to leading web application development frameworks like R Shiny, Bokeh, Flask, and Javascript. Learn more about applications in Dataiku here. Dataiku 9 adds the Dash framework, one of the most popular application development frameworks on the market. With all of the leading frameworks now integrated with Dataiku, teams can choose the best development tools for their needs.
What-If Analysis for Data and Business Teams
The explainability of predictive models is critical to building trust in model outputs. What-if analysis (through the interactive scoring capability) in Dataiku 9 allows data scientists to input real or made up data values and see how the model behaves. Even better, data scientists can publish what-if analysis to a dashboard where a business analyst or domain expert can also try different values and see the response. This ability to visually test assumptions and see responses helps everyone working on AI projects build trust and understanding in model outputs. Learn even more about what-if analysis in the video at the end of this post.
What-if analysis in Dataiku 9