Case Studies

AI is Versatile . . . and so is

Berrijam AI empowers business teams with data analytics “super-powers.”

Goal: Understand Heart Attack mortality

Why: Accurately triaging patients and reducing time to treatment can save lives.

What we found: Berrijam AI's analysis of ICU data on heart failure patients identified Anion Gap and Rel Failure as the top mortality indicators, increasing the mortality risk by ~2x.

Industry: ICU | Blood-works | Medical | Research

Summary

View full discoveries

Try it out yourself

About the data:

Data sourced from Datadryad

Preparation: We added column 'Mortality' based on outcome value.

Goal: Understand the factors leading to customer churn

Why: Customer satisfaction and churn are important when it comes to retaining customers and growing a business. Understanding why some some customers are leaving, and designing targeted interventions can help business grow.

What we found: Customers on Fiber Optics service for a US Telco had a 85% higher risk of leaving

Industry: Telecommunication | Customer Retention

Summary

View full discoveries

Try it out yourself

About the data:

Data sourced from Maven Analytics Churn Challenge

Preparation: We added a column 'Churn', where churned customers were set to 'yes' and stayed customers were set to 'no'.

Goal: Understand the factors of highly innovative regions in Australia

Why: Insights can help guide strategic initiatives and development of policies to foster innovation.

What we found: Only 7 of the 40 factors explored were predictive of innovative regions, reflecting that most innovation in Australia taking place in major cities south of Sunshine Coast and in areas with fewer businesses in Agriculture, Forestry and Fishing industries

Industry: Strategy | Government Policy | Innovation

Summary

View full discoveries

Try it out yourself

About the data:

Data sourced from Department of Industries, Science and Resources (DISR)

Preparation: We defined a column 'Highly Innovative', where a region had a value of 'yes', if the number of patent applied from the region were within the top 20% of number of patent applications.

Goal: Understand which combination of factors result in machine failure.

Why: Predicting machine failure with high accuracy can significantly reduce downtime and maintenance costs in industrial settings.

What we found: You can reduce 50% of machine failures by proactively monitoring sensor readings.

Industry: Manufacturing | Predictive Maintenance

Summary

View full discoveries