Experimental statistic

Estimates of modalities of use of websites by enterprises

  • Ascolta questa pagina usando ReadSpeaker
  • Condividi
  • Lascia un feedback

Istat makes available the estimates referred to year 2018 about functionalities and services offered by the enterprises’ websites. The information collected by the survey on ICT use of the enterprises with 10 or more persons employed and the information directly collected from the enterprises’ web pages, supports the estimation procedure according to the following process:

1 – Web address acquisition URL from the admin sources
URL from thematic directory sites
URL from batch queries on search engines (URL Retrieval techniques in case of non existing URL)
2 – Enterprise identification URL validation, check URL’s validity (recurring errors and domain extraction)
Detection of identification variables from the website and comparison with the same information available in the SBR register
3 – Data analytics Web Scraping techniques for web data acquisition
Text Mining techniques for extracting the requested information
Machine Learning techniques for the use of algorithms that simulate a learning process for the construction of predictive models
4 – Inference From the enterprises with scraped websites to the enterprises of the target population

(The main phases for estimating the target distributions that uses information from the website)

In download, in Excel format, the estimates of the rate of enterprises (on the total reference population) which:

  1. offer web ordering functions;
  2. propose job offers or gives information on job vacancy in the enterprises;
  3. have links to social media (Facebook, Twitter, Instagram etc.);
  4. show a combinations of some functions and/or services on the website (statistics on the discrepancy between the current estimates and the estimates using information from the web sites).

Compared to the edition referring to 2017, the 2018 estimates present three main novelties: an update of the data mining process, a modified estimator to reduce the selection bias related to the Big Data source; the addition of the parameter estimates defined in point (4).

FOR FURTHER INFORMATION

  1. Technical Report
  2. Producing contingency table estimates integrating survey data and Big Data (G. Bianchi, G. Barcaroli, P. Righi, M.Rinaldi)

Reference period: Year 2018

Date of Issue: 08 May 2020

Experimental statistics

experimental statistics Do you know what experimental statistics are?

FEEDBACK
You are invited to leave your observations, comments and suggestions on our experimental statistics.
Please email to: statistiche-sperimentali@istat.it