Automatic Data Science: DataRobot, Quill and Loom Systems
Automatic data science is on the rise. This post examines three recent visionary solutions: DataRobot, Quill and Loom Systems.
Quill
Developed by Narrative Science, Quill ingests raw data and outputs it as conversational language reports. By doing this automatically, it enables data-driven communications to take place at enterprise-scale.
Quill is based on Natural Language Generation (NLG), a process which conveys the most relevant insight found in raw data through spoken language. It’s powered by artificial intelligence, enabling it to analyze data from disparate sources, understand hidden relationships and transform these into coherent narratives.
While the vast majority of analytics solutions turn to data visualization, Narrative Science is breaking the mold by telling users the story behind their data. By providing them with clearly written, narrative-driven results.
Quill can produce reports about virtually any data-driven topic. The reports can vary in length, tone, and style, and are customizable to their intended audience.
Platform administrators input their business rules, which are used to identify thresholds, trends and relationships in the data. This enables Quill to determine the most relevant data for each type of report.
Features and Capabilities
- Predictive Analytics: No, Quill ingests raw data and reports on it, but does not perform trend projections
- Automatic: Partial, some of the data streaming into the platform needs to be manually pre-processed
- Data types: Structured only, in any of the prevalent formats (JSON, XML, CSV, etc.)
- Natural Language Generation: Yes
- Professional Training: None required
- Transparency: No, the analysis takes place under the hood.
- Data Visualization: No
- Delivery Model: Cloud and on-Premise
- Implementation Time: 2 to 12 weeks
Loom Systems
Loom Systems simplifies data scientists’ work cycle by fully automating a large part of it. The platform automatically ingests and analyzes all types of structured and unstructured data, both batched and streaming. It monitors the data 24/7, performing correlations between all data sets to find anomalies, outliers and trends.
Parameters, baselines and thresholds are set automatically as soon as the system goes up, and are continuously optimized according to its unique behavior.
Incidents in Loom Systems are accompanied by an automatically identified root cause, as well as recommended resolutions from a proprietary resolutions database. Platform users also fill in their own suggestions for remediation, next steps and resolutions, which are presented whenever incidents recur. This helps retain organizational knowledge.
Loom Systems was built with the explicit intent of being a low-touch, operationally simple and usable solution designed to help organizations extract insight from their data fully automatically. However, as it is aimed at technical professionals, it still allows users to drilldown into their raw data and perform deeper investigations.
Features and Capabilities
- Predictive Analytics: Yes
- Automatic: Fully automatic from start to finish, improving partially through user feedback input but mostly through adaptive machine learning
- Data types: Structured and unstructured
- Natural Language Generation: Yes, concisely
- Professional Training: None required
- Transparency: Partial, a large part of the analysis takes place under the hood, though users can deep-dive directly into the relevant raw data from within the UI and perform as detailed an investigation as they choose.
- Data Visualization: Yes, as a supporting feature
- Delivery Model: Cloud and on-Premise
- Implementation Time: 90 minutes
- Who Is It Right For: Loom Systems is the go-to solution fast moving businesses who prefer to fully automate their big data analytics. Quick access to incident root cause, insights and recommended resolutions as well as a robust knowledge retention system.
Bio: Beerit Goldfarb is Director, Technical Community and Evangelism at Loom Systems.
Related:
