forecasting<\/a> is not an exact science and relies on several truths and assumptions. These being:<\/p>\nAssumptions that historical trends and patterns shall continue to replicate with varying degrees over time.Forecasting will contain errors and anomalies because your data set (your real-world clicks data) will contain anomalies that could be construed as errors.Forecasts typically revolve around the average, making group forecasts more reliable than running a series of micro-forecasts.Shorter-range forecasting is typically more accurate than longer-range forecasting.<\/p>\n
With this out of the way, we can begin to model and forecast our traffic data.<\/p>\n
For this article, I will visualize our data as a Bayesian Structural Time Series (BSTS) forecast, one of the packages we installed earlier. This graph is used by most forecasting methods.\u00a0<\/p>\n
Most marketers will have seen or at least be familiar with the model as it is commonly used across many industries for forecasting purposes.<\/p>\n
The first command we need to run is to make our data fit the BSTS model:<\/p>\n
ss <- AddLocalLinearTrend(list(), ts_data$Clicks)
ss <- AddSeasonal(ss, ts_data$Clicks, nseasons = 52)
model1 <- bsts(ts_data$Clicks,
state.specification = ss,
niter = 500)<\/p>\n
And then plot the model components:<\/p>\n
plot(model1, “comp”)<\/p>\n
And now we can visualize one- and two-year forecasts.<\/p>\n
Going back to the previously mentioned general forecasting rules, the further into the future you forecast, the less accurate it becomes. Thus, I stick to two years when doing this.<\/p>\n
And as BSTS considers an upper and lower bound, it also becomes pretty pointless past a certain point.<\/p>\n
The below command will produce a one-year future BSTS forecast for your data:<\/p>\n
# 1-year
pred1 <- predict(model1, horizon = 365)
plot(pred1, plot.original = 200)<\/p>\n
And you\u2019ll return a graph like this:<\/p>\n
\n<\/div>\n
To produce a two-year forecasting graph from your data, you want to run the below command:<\/p>\n
pred2 <- predict(model1, horizon = 365*2)
plot(pred2, plot.original = 365)<\/p>\n
And this will produce a graph like this:<\/p>\n
As you can see, the upper and lower bounds in the one-year forecast had a range of -50 to +150, whereas the 2-year forecast has -200 to +600.<\/p>\n
The further into the future you forecast, the greater this range becomes and, in my opinion, the less useful the forecast becomes.<\/p>\n
The post How to use RStudio to create traffic forecasting models<\/a> appeared first on Search Engine Land<\/a>.<\/p>","protected":false},"excerpt":{"rendered":"There is a lot of fervor in the SEO industry for Python right now. It is a comparably easier programming language to learn and has become accessible to the SEO community through guides and blogs. But if you want to learn a new language for analyzing and visualizing your search data, consider looking into R. This article covers the basics of how you can produce time series forecasts in RStudio from your Google Search Console click data. But first, what is R? R is \u201ca language […]<\/p>\n","protected":false},"author":0,"featured_media":0,"comment_status":"open","ping_status":"open","sticky":false,"template":"","format":"standard","meta":{"h5ap_radio_sources":[],"cybocfi_hide_featured_image":"","footnotes":""},"categories":[1],"tags":[],"class_list":["post-7107","post","type-post","status-publish","format-standard","hentry","category-best-seo-blogs"],"featured_image_src":null,"featured_image_src_square":null,"author_info":{"display_name":false,"author_link":"https:\/\/craighordlow.com\/author\/"},"_links":{"self":[{"href":"https:\/\/craighordlow.com\/index.php\/wp-json\/wp\/v2\/posts\/7107","targetHints":{"allow":["GET"]}}],"collection":[{"href":"https:\/\/craighordlow.com\/index.php\/wp-json\/wp\/v2\/posts"}],"about":[{"href":"https:\/\/craighordlow.com\/index.php\/wp-json\/wp\/v2\/types\/post"}],"replies":[{"embeddable":true,"href":"https:\/\/craighordlow.com\/index.php\/wp-json\/wp\/v2\/comments?post=7107"}],"version-history":[{"count":0,"href":"https:\/\/craighordlow.com\/index.php\/wp-json\/wp\/v2\/posts\/7107\/revisions"}],"wp:attachment":[{"href":"https:\/\/craighordlow.com\/index.php\/wp-json\/wp\/v2\/media?parent=7107"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"https:\/\/craighordlow.com\/index.php\/wp-json\/wp\/v2\/categories?post=7107"},{"taxonomy":"post_tag","embeddable":true,"href":"https:\/\/craighordlow.com\/index.php\/wp-json\/wp\/v2\/tags?post=7107"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}