Briefing

Situation

Did you know that Istanbul alone devours a staggering 2,000 tons of baklava during the festive season? That’s a lot of sweet treats!

However, with great treats come great responsibilities!

  • Two years ago, the country produced way too much baklava for the demand

    • This nearly bankrupted Türkiye, and greatly affected driving conditions as pallets of baklava had to be stored in the streets.

  • Last year, in an attempt to avoid a repeat, the government massively underestimated the demand, as people rushed to stock up for 10 years. This led to massive disappointments throughout the country.

  • This year, to ensure that every baklava lover gets their sweet fix, and to meet this incredible demand, a predictive AI model was built to predict the demand as accurately as possible! The country’s future hangs in the balance!

Mission

But Türkiye believes it can do better and has hired you to help!

Let’s work together to ensure a successful festive season for everyone!

We invite you to join our effort to optimize this model further. Your task is to:

  • Experiment with Hyperparameters: Tweak parameters like epochs, batch_size, learning_rate, and hidden_layer_units to find the optimal configuration.

  • Store Your Experiments: Record the hyperparameters you used, the resulting model artifacts (the trained model), and evaluation metrics in Model Registry.

  • Deploy the Best Model: Once you’ve identified the most performant model, deploy it to your environment from the Model Registry.

  • Test Your Model: Use the provided notebook to make predictions and evaluate the deployed model’s performance.

By making your experiments and storing your findings in Model Registry, you’ll be part of a collaborative effort to make the most of this sweet AI challenge.