Section: 15 🔖 Free Style: Prediction

• Lecture slides:

15.1 Snippet 1: Example of a simple freestyle prediction model

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

15.2 Snippet 2: How to build a freestyle (your own code) prediction model?

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

15.3 Snippet 3: One-step crossvalidation

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

15.4 Snippet 4: Preparing submission.csv for Kaggle

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