Business Case: Data provision for AI forecasts
Sort large amounts of data and make them usable
An AI company equips the retail trade with analytical data on storage capacity and demand forecasts. The large amounts of data are a valuable by-product - but only if someone cleanses it, improves it and makes it available for analysis. This is where we come in.
© Oliver Hallmann: Äpfel, Licence CC BY 2.0
When do customers want to buy this fruit? Artificial intelligence can be used to predict that. And make a lot more out of the resulting data.
Project requirements/tasks
Our customer has developed software that enables retailers to improve their delivery and warehouse processes. To do this, our customer receives raw data from the markets and uses it to develop complex forecasts. This is another added value he offers retailers: It makes optimized data sets available for their own analyses - a double service. To do this, however, data volumes in the terabyte range have to be structured and cleaned up. Additional information such as weather data or relevant events can also be added: If you know when a football final is due to take place in the summer heat, you can offer cold drinks that are in high demand.
This is our contribution
As data engineers with a knowledgeable view of our customer's data science activities, we enrich the resulting datasets and make them available. No trivial task, because extracting and supplying data sets of this size requires a partner with a high level of big-data competence.
Thanks to our contribution, our customer is not only able to supply retailers with its AI forecasts. He goes the extra mile because we also provide him with high-quality data sets that we have formed from the huge amounts of raw data and thus made ready for use.
The project at a glance
-
Project title:
-
Data preparation and provision for AI forecasts
Industry:
-
KI-Software
Term:
-
6 months
-
-
Our contribution:
- Data cleansing
- Data enrichment
- Provision of data
Technologies:
- Microsoft Azure
- Python Dask
- Python Pandas