We are excited to announce the latest release for Alteryx Machine Learning. This release comes with exciting new Time Series enhancements that will dramatically expand the predictive power of our flagship machine learning product. This release also features a user interface (UI) update with new model evaluation capabilities making the process of model development more intuitive and easier than ever.
Since its launch, Alteryx Machine Learning has proven to be a powerful tool for machine learning upskilling for analysts, advanced analysts, and LOB knowledge workers, making it a key piece of our Analytics Cloud Platform. It does this through an intuitive and easy-to-use, cloud-native AutoML product, complete with an education mode that teaches advanced analytics methodology and best practices throughout the process. By also bringing fully-guided AutoML using best-in-class models, Alteryx Machine Learning optimizes time-to-value for model creation while it teaches.
Our current product provides a robust solution for reducing the workload on data science teams and enabling LOB analysts in finance, human resources, and marketing departments to apply machine learning to their unique business problems. The office of finance for our customer, eBay, uses machine learning for tax classification of products, enabling them to reduce the amount of manual work required for reporting and audit purposes by 50%. In a short amount of time, Alteryx can help customers go from demo, to model creation, to scaling data science across the organization.
Maximum visibility and ease-of-use for model creation
The interface for Alteryx Machine Learning was designed around making each step of the AutoML process clear, visible, and intuitive. Each step of this process is fully guided with a clear tabbed format for easy navigation between the steps. With this release, we are launching a new tab called Model Setup which will help the user understand all the components that go into creating the model and to provide better contextualization. When time series regression models are selected, this model setup step will clearly show a decomposition the series components.
In this view, users can easily visualize a useful abstract model of the data, giving clear insight into any problems present, as well as giving quick conceptual understanding and explainability of trends and seasonality. Present in this view, are:
- The observed time series data, which is simply a direct plot of the observed values over a specific period
- Seasonality in their data to account for things like holiday shopping rush, or activity differences between the work week and weekend
- Longer-term upwards or downwards trends separated from short-term fluctuations
- The residuals, which gives a view of how well a model fits data and enables optimization using non time series models
Model Evaluation and Operationalization made easy
Once a model is created, Machine Learning simplifies the steps of getting it into production. For time series model evaluation, this includes automatically testing model performance against holdout data and providing a clear view into how the model compares to observed data.
This enables easy tuning and correction of models all within the same platform. Users can then export Time Series Forecast Graphs, enabling users to easily visualize and communicate forecasts. This includes observed trend data and the predicted values.
Book a Time Series Demo Today
Alteryx Machine Learning has transformed how organizations develop and operationalize machine learning models by bringing the capabilities, the methods, and the models to the business user. This takes ML projects off overloaded data science teams and enables the workers with the industry and business knowledge to uncover answers to their most pressing business problems.
Current customers can test out the new UI changes and time series enhancements today but to get a peek into how Alteryx can be a primary driver of DSML upskilling for your org while helping you develop impactful machine learning models, schedule a demo today.