Perion is an ad-tech company who delivers data-driven online advertising and search solutions to brands and publishers. Perion helps brands capture consumer attention via beautiful creative and
proprietary ad formats served across top-tier publishers. The goal of this project is to explore the essence of creative and its impact on performance for different audience groups.
- Long time running due to big amount of data
- Find the bests features to use
- Unbalanced data
- Find the best model and metrics to use
- Hyperparameters tuning
Achievements (according to KPIs)
- Export the data using PostgreSQL
- Disaggregate the data
- Create adaptive models for each ad sets
- Ensemble those models
- Trained and tested this model to find the best hyperparameters and the performances
- Make predictions
Dan received this assignment and in short time, with a lot of dedication and creativity in mind, had made an impact on how the company is approaching its creative data. The results of his work were presented to senior management in Perion and was received with great appreciation and excitement, which led to new directions and new research opportunities.
Dan had to learn in an auto-didactic manner, technologies he had little experience in (Spark) and work on big and semi-structured data. He also fit well in the team and made good connections with peers.
Dan’s work is just the tip of the iceberg in terms of future research in an area the company had not touched before. Future development is now very relevant and will be discussed with the management.