Hello
For Wrap-up quiz 7, question 4, are you sure you have a mean absolute error of 44. I got an average error of 64.3
Please can you maintain 44 ?
Many thanks.
Hello
For Wrap-up quiz 7, question 4, are you sure you have a mean absolute error of 44. I got an average error of 64.3
Please can you maintain 44 ?
Many thanks.
I confirm the answer provided by the correction. Running the code with 10 different random state, I get the following results:
MAE on test sets:
[44.71263363 43.9783554 42.92460634 43.10264111]
mean +/- std: 43.680 +/- 0.718 Watts
MAE on test sets:
[44.72913069 44.22828706 43.78499244 43.34948918]
mean +/- std: 44.023 +/- 0.513 Watts
MAE on test sets:
[44.4079987 44.02684056 44.04079353 44.02977772]
mean +/- std: 44.126 +/- 0.163 Watts
MAE on test sets:
[44.38554846 43.63357771 44.06888205 43.36563006]
mean +/- std: 43.863 +/- 0.392 Watts
MAE on test sets:
[44.27589829 43.37409993 43.5017239 43.38746364]
mean +/- std: 43.635 +/- 0.373 Watts
MAE on test sets:
[44.35690354 44.2235122 43.96400091 43.36537854]
mean +/- std: 43.977 +/- 0.381 Watts
MAE on test sets:
[44.217028 43.63513624 43.5672384 43.58098014]
mean +/- std: 43.750 +/- 0.271 Watts
MAE on test sets:
[44.21414606 43.78332638 43.85198492 43.61353516]
mean +/- std: 43.866 +/- 0.219 Watts
MAE on test sets:
[44.42887602 44.44177659 43.65024002 43.30972684]
mean +/- std: 43.958 +/- 0.493 Watts
MAE on test sets:
[44.50207271 44.27864666 43.75761911 43.44140245]
mean +/- std: 43.995 +/- 0.418 Watts
OK thanks, I understand my mistake.
It is solved.
Hi;
I’ve run this code in the “Show Answers” in a “sandbox” notebook and still do not get the same answers as you do. I get the same average error as
sdagoret.
I run the same code on the notebook in my computer and get the same answer. MEA of around 65.
From this thread Explaination for Question no.8 - #5 by saurabhmishra, I might infer that you are using the entire dataset instead of the hand-crafted dataset that was used to build the model based on physics.
By using parameter such as hear-rate, the model will make less error but it is not based on the physic anymore.