Expert Models

Production-quality data science requires a major team effort. The good news is our Expert Models platform takes care of a large number of jobs required to build successful data science projects and products.

Summary

Expert Models is our sophisticated, yet user-friendly data-science platform which enables you to collaborate, share data and create and deploy models online. We have the data science platform and technology expertise you need and we are here to help.

Request A Demo or Start a Conversation Today

Advantage of our platform

Save time. get large scale projects into production and to market in a small number of weeks (can be < 4 – 6 weeks) vs. a large number of months (normally 12+).

Save recruitment and staff costs. We take care of your platform, maintenance, development operations, systems, middleware programming (user management, data management, computing management), interface design and implementation and quality assurance of the platform.You don’t need to buy traditional expensive enterprise systems to get world-class production-quality.We can also collaborate with you on the data science model and data development and analysis if your team is already at capacity.

Get to the market faster. You don’t need to spend months installing and configuring heavy duty traditional expensive enterprise systems anymore. You can be up and running in minutes.
Take advantage of Python. Staff and consultants with the obscure skills required to manage and develop on traditional enterprise systems are rare and expensive. Expert Models is built on Python with well documented APIs. The supply of Python expertise is well established and growing rapidly (e.g. Python is even being taught in business schools now). You own the IP to your models that you develop (or that we develop for you under contract) and you can reuse them or redeploy them on any compatible system as you require. Tap into a large range of libraries and support – Python has a large number of free to use and powerful data science libraries.

Integrate with your existing system. EM is compatible and can establish connectivity to a wide range of IT technology infrastructures. It has standard APIs that can be used for almost any existing IT infrastructure (no matter what technology was used).

Centralise all of your data and code. Expert Models can import CSV files using HTTP/ FTP if APIs are not available.

Enable better teamwork. Everyone on the team can see and work on the latest version of the data, code and analyses in one secure, central location. Version control and backups of all updates are included.

Prototype quickly. Get analytics prototype models into production and in front of your partners and stakeholders in minutes or hours rather than weeks. This allows you to get feedback and iterate quickly with your partners or clients.

Reduce operational risk. Expert Models tracks, version controls and snapshots every version of data and model code used in your analytics so that all of your historic analyses are stored and reproducible.

Satisfy regulation. Expert Models tracks, version controls and backs up every version of data and model code so that you will be able to satisfy authorities that you can stand over your decisions and data analytics.

Demo

The Interface

The default Expert Models setup consists of an advanced file management system including data sharing (similar to Google Drive) but designed specifically for data science. File organisation is an essential workflow to get right for anyone working with data.

Secondly, Expert Models has a data editor . This can be thought of as a simplified version of Excel, but with features useful for working with the type of large datasets used by models. Data table types can defined to ensure data is correct at entry, rather than runtime.

The End User of Expert Models

The modeller: Most models are not online and do not live beyond contributing to a report. These models can offer great value if made available to other audiences. One of the reasons this has not happened already is that Modellers, while they may have advanced modelling skills, are not web programmers. Expert Models allows Modellers host their models within a web interface without having to learn web programming / system administration, or engage in a long expensive development project.

We work with Modellers to publish their models online in an interface we agree is appropriate. The Modeller can then open these models up to new audiences.

The Excel user: Excel still dominates the workflow of Data Analysts and Decision Makers alike. As well as Excel being used for its obvious function of working with data, Excel is also the primary go to software for creating mini-programs that perform some repeatable action. These programs can include models, data-cleaners, connectors and other variations. All these types of program are better implemented in Expert Models. Excel has major downsides that are increasingly becoming unacceptable in a configured world. A recent IBM report focused on just this issue:

“A number of studies have been made concerning the frequency of errors in spreadsheets. Based on these, it seems that 90 percent of all sheets contain at least one error.”

These errors can be avoided when a program is built within a workflow that includes data validation and programmatic error checking mechanisms. Building models in Python inside Expert Models will inherently clean up errors.

Scripting data import from your various data sources in these Python blocks, inside the platform, also has the double effect of allowing you to understand and audit the import process and remove this repetitive task.

Getting up and Running

Experience has told us that all customers will need some kind of support and sometimes customisation to their solution that goes beyond the default Expert Models setup. For this reason we not only provide a turn-key solution like CRM packages where the problems are already very well defined, we also provide a team of experts who will work with you to understand your problem, then we help you get your models and data into Expert Models before handing over logins.

Are you interested in trying a demo account? If so, please contact us.