The Democratization of Predictive Analytics
How does it work?
Business cases
Success Stories
Leading medical company
Revenue forecasting by 2000+ SKUs and 80 regions with automatic selection of optimal model for each forecasting unit and options to include additional factors to improve accuracy
CHALLENGES
Large number of unique SKUs that need to be forecasted separately
Limited availability of data science specialists
Forecast accuracy issues
BENEFITS
Ability to calculate forecasts by each SKU and region
Support for managers in forecasting activities
Ability to add new factors and variables into forecasting process
Integration with Anaplan models
Leading internet company
Sales forecasting by 2 million of B2B customers taking into account internal and external attributes
CHALLENGES
No statistics indication for sales managers in forecasting
Judgmental estimations errors
No data science specialists are available for particular forecasting tasks
BENEFITS
Ability to calculate forecasts by millions of existing customers
Ability to provide basis for forecasting of new customers
Challenge for judgmental forecasting
Ability to add new factors and variables into forecasting process
Leading telecom company
Content recommendation service for each user based on consumption history and matching similar combinations of multiple content characteristics
CHALLENGES
Slow process of selecting recommendations using 50 parameters and 400 content characteristics
Hours of work to manage machine learning algorithms
Low data science specialists availability
BENEFITS
Confirmed the ability to tune machine learning algorithms without any programming by analysts
Increased prediction accuracy of ROC AUC from 72% to 86%
Accelerated machine learning models tuning by 2 times