| Popis: |
Nowadays, companies wishing to get exposure and increase revenues need Internet visibility, and particularly get on the top of Search Engines Results Pages (SERP). Meanwhile, in the last two decades, data science became a vital part of many industries to get quick insights and valuable information, including online marketing and Search Engine Optimization (SEO). With 90% market share worldwide, Google is a search engine using an algorithm to query billions of pages with more than 200 factors that all marketers want to discover. As consequence, some applications are existing to get insights about SEO factors relative to webpages. However, it does not exist a tool able to collect large quantities of URLs and analyze them relative to on-page factors. Therefore, a custom tool "SEOnaozi" was developed for this purpose. This article tries different machine learning algorithms to evaluate the SEO tool and determine how effective it can be to improve SEO analysis. For this purpose, three datasets for a total of 50,000 observations were tested with two scenarios: one using a selection of 8 features, the other with 23 features, including 15 on-pages features. According to the research result, it leads that the tool "SEOnaozi" is efficient because all metrics mostly improved when on-pages features are used by algorithms. Moreover, it was found that Catboost could produce the best results among others to predict an URL to be on Google Top10 SERP. |
