HW1 Q&A and Discussion by Ed Saiedi https://padlet.com/ed_saiedi1/vmp1miah6x0iu7ua Write any questions you have here and we'll try to answer. We'll also include answers to questions emailed here for all to see. Do check if a question you may have has been answered here. en-us 2023-02-22 00:28:46 UTC 2023-03-07 12:10:09 UTC hello@padlet.com Question about HW1: It's a little bit confusing in the excel part. There is a "SalePrice" described as "Cost of sale" in the dictionary. meichen8_1 https://padlet.com/ed_saiedi1/vmp1miah6x0iu7ua/wish/2506771150 Answer: "SalePrice" means that the sale price of a tractor, becomes the cost of the sale for the other party monetarily (it's a clumsy wording by providers of the data, that is better formulated as the purchase cost, rather than the cost of sale. )]]> 2023-03-07 11:56:37 UTC https://padlet.com/ed_saiedi1/vmp1miah6x0iu7ua/wish/2506771150 Question about HW1: Since the data is "historical data", do I record that 100 000 tractors were sold? meichen8_1 https://padlet.com/ed_saiedi1/vmp1miah6x0iu7ua/wish/2506781935 Answer: The best (and most correct) method for doing the homework, is not to calculate the variables you have for all 100,000 tractors, but a certain portion of it, for which you could have both the performance of the baseline model AND the out-of-sample performance of the ML model. Remember that you wish to compare the ML model with the baseline model. For a correct comparison, you can't make that comparison on the whole data, but you must restrict it to a portion. I cannot tell you what portion that is, as I would give the answer away. 
 

]]> 2023-03-07 12:05:39 UTC https://padlet.com/ed_saiedi1/vmp1miah6x0iu7ua/wish/2506781935