Reader small image

You're reading from  Forecasting Time Series Data with Facebook Prophet

Product typeBook
Published inMar 2021
Reading LevelBeginner
PublisherPackt
ISBN-139781800568532
Edition1st Edition
Languages
Tools
Right arrow
Author (1)
Greg Rafferty
Greg Rafferty
author image
Greg Rafferty

Greg Rafferty is a data scientist in San Francisco, California. With over a decade of experience, he has worked with many of the top firms in tech, including Google, Facebook, and IBM. Greg has been an instructor in business analytics on Coursera and has led face-to-face workshops with industry professionals in data science and analytics. With both an MBA and a degree in engineering, he is able to work across the spectrum of data science and communicate with both technical experts and non-technical consumers of data alike.
Read more about Greg Rafferty

Right arrow

Correcting outliers that cause seasonality swings

We'll be using a new dataset in this chapter to look at outliers: the average number of likes per day of posts on National Geographic's Instagram account, @NatGeo. This data was collected on November 21, 2019.

I've chosen this dataset because it exhibits several significant outliers, which are marked in the following plot:

Figure 9.1 – Outliers on National Geographic's Instagram account

Each dashed vertical line indicates a moment where the time series deviated significantly. The second line from the left indicates a radical trend change in the summer of 2015 but the other four lines indicate outliers, with the last two outliers spanning across wide time ranges. We'll specifically be looking at the line occurring in mid-2016, in August to be precise. This represents the most extreme outliers. The 2014 set of outliers can be safely ignored, as they do not affect the forecast...

lock icon
The rest of the page is locked
Previous PageNext Page
You have been reading a chapter from
Forecasting Time Series Data with Facebook Prophet
Published in: Mar 2021Publisher: PacktISBN-13: 9781800568532

Author (1)

author image
Greg Rafferty

Greg Rafferty is a data scientist in San Francisco, California. With over a decade of experience, he has worked with many of the top firms in tech, including Google, Facebook, and IBM. Greg has been an instructor in business analytics on Coursera and has led face-to-face workshops with industry professionals in data science and analytics. With both an MBA and a degree in engineering, he is able to work across the spectrum of data science and communicate with both technical experts and non-technical consumers of data alike.
Read more about Greg Rafferty