Sprint #9 (June 29)

[pagand]

Feedback:

• Explore the possibility of grouping the dataset into distinct clusters and train multiple models to enhance accuracy.
• Use r2 score to evaluate the model's performance
• Correct the decoder for categorical variables: decoding based on a base of 2 and scaling by dividing the maximum value.
• Change the hyper-parameters (horizon and seq length to obtain better result)
• Tweak model features (wind, different weather feature) to see the best result in specific model
• Consider transformers pertained model
• Make your test set isolated for consistent comparison

Action items:

• Draft paper (prepossessing stage)
• Develop a model with an R2 score exceeding 0.99

[yim-fan]

• [x] Draft paper (prepossessing stage)
  • First draft finished
• [x] Develop a model with an R2 score exceeding 0.99
  • In process with the new output and input