Automation of model predictions with Machine Learning in the cloud, requiring less preparation time and improved speed to market
This case study is an example of our solution forMachine Learning
The analytics team was processing numerous new requests to generate new predictive models and support the business with insights. The idea was for us to work together to create a readily available virtual workspace for Machine Learning analysis with model generation powered by the datasets and training sets that could be easily ingested in order to proceed with model development and preparation time.
As a partner to SickKids foundation, Bitstrapped worked to advance the culture of discovery and insight through the automation of machine learning. Taking advantage of managed services in the cloud, a solution was implemented to generate new models with feature engineering and explainable AI in minutes, which previously took months. This enabled the team to generate models with SQL to pre-process, analyze, train and deploy their models, all within a familiar SQL interface.
Initial results indicate that new model performance is comparable to prior more manually intensive processes. The analytics team was able to reduce the time to complete and iterate on models, while testing multiple models at once with varied parameters. The team's ability to quickly experiment was ramped up due to the delivery results based on using SQL queries and templated workflows. As a benefit to the business, there was a shift from majority of time preparing, to majority of time analyzing, and increased productivity.