[QUESTION] Why is your implementation not using the previous curve

[Vizepi]

In the mikebryant's implementation and according to the initial data analysis by Ninji, you can deduce the curve type from the previous curve type (varying/small spike/big spike/decreasing).
Why is this information not required by your implementation? Is it smart enough to detect the curve type with less data?
I'm curious :-)

[elxris]

Previous week data only adjust probabilities results, but still, for any previous week, the probabilities are non zero.
So, in a way, this calculator is more dumb. I have plans to calculate and give the fine-grained probabilities number, so I will ask for previous patterns in the following updates.

[TheJus10]

Previous week data only adjust probabilities results, but still, for any previous week, the probabilities are non zero.
So, in a way, this calculator is more dumb. I have plans to calculate and give the fine-grained probabilities number, so I will ask for previous patterns in the following updates.

Is there any place to report actual data? I assume (or hope) that the calculator doesn't send live-updates if it does send any data at all. Do you even need data reported at this point or is the system entirely algorithm based so additional data is not necessary?

[elxris]

I meant that I will ask if the user knows what pattern was in the previous week so the user could input it. The calculator could give more precise probabilities if it knows the previous pattern. I don’t send data to servers or something, it’s not necessary.