Vol. 2: Embracing the Unknown – Exploring Data Science Thesis at #Otemachi

It was a pleasure to join the study group today to read and discuss The Prophet thesis. I am grateful to Mr. Kayamoto for leading such a fascinating session and to everyone who participated. It was exciting to explore how to incorporate formulas into the Prophet Forecast Model using Excel. Our goal for today was to create a growth forecast for user numbers and population trends based on historical data in Excel.

In the thesis, I learned that the decomposable time series model consists of three main components: trend (g(t)), seasonality (s(t)), holidays (h(t)) and error (εt) as illustrated in the first fomula below. The third formula is exactly the one which puzzled us because it's not so important to understand the structure of the model to incorporate change points in the trend. The fourth formula is good to understand the incorporation of the change points.

Thank you all for your engagement and participation.

reference: https://peerj.com/preprints/3190/

#1

#2

#3

#4