Post by account_disabled on Feb 20, 2024 1:08:12 GMT -6
Important vital events such as the Corona virus undoubtedly affect internet usage behavior. After the announcement of the first case in Turkey, everyone was wondering what the economic impact would be. In addition, the behavior (or reflection of the psychology) of internet users the day after such an announcement is made and a few days after the announcement is made is also a matter of curiosity for me. A few days after the announcement, I decided to analyze the traffic of several websites. At the same time, I wanted to understand the sectoral effects of such events by categorizing these sites.
For this analysis, I used time series forecasting models. Here, to speed Greece Phone Number up the analysis process, I used Facebook's modeling package called "Prophet" instead of ARIMA type models and modeled the number of sessions of more than 50 websites (and shared some of them in this article). In order not to share real traffic, I will normalize the data of each site and show it. Below you can find data from one of the normalized and modeled sites: The red line is the normalized traffic values of an e-commerce site. (before March 11, 2020) The blue line is the traffic change according to the model created by Prophet.
The greenish line is traffic data after March 11, 2020. In this way, we can clearly see that if there had been no announcement about the coronavirus, the decrease in the traffic of the above e-commerce site would not have occurred. It is also possible to see from here that the traffic of the site has reached its lowest levels this year. Another example: Facebook's prophet time series forecasting has created a much more successful model in this example. In this example, we see that the day after the announcement, users most likely visited this site to check their education status. We can also say that traffic returned to almost normal levels after the sixth day.
For this analysis, I used time series forecasting models. Here, to speed Greece Phone Number up the analysis process, I used Facebook's modeling package called "Prophet" instead of ARIMA type models and modeled the number of sessions of more than 50 websites (and shared some of them in this article). In order not to share real traffic, I will normalize the data of each site and show it. Below you can find data from one of the normalized and modeled sites: The red line is the normalized traffic values of an e-commerce site. (before March 11, 2020) The blue line is the traffic change according to the model created by Prophet.
The greenish line is traffic data after March 11, 2020. In this way, we can clearly see that if there had been no announcement about the coronavirus, the decrease in the traffic of the above e-commerce site would not have occurred. It is also possible to see from here that the traffic of the site has reached its lowest levels this year. Another example: Facebook's prophet time series forecasting has created a much more successful model in this example. In this example, we see that the day after the announcement, users most likely visited this site to check their education status. We can also say that traffic returned to almost normal levels after the sixth day.