As for the USA Presidential Forecast I used my algorithm with some (better: many) needed changes to adapt it to the Italian Referendum scenario.
Today, it is not possible to publish the results due to the Italian laws about referendum but I created a .png file that shows them. Then I obtained an on-line timestamp by http://www.originstamp.org/ and so I will be able to publish the .png forecast file tomorrow proving exactly when I created it.
Moreover, I published a Tweet with the file hash of that .png file in order to publicly prove it:
“…Cambridge Analytica, has been using Facebook as a tool to build psychological profiles that represent some 230 million adult Americans…
…No data point is very informative on its own, but profiling voters, says Cambridge Analytica, is like baking a cake. “It’s the sum of the ingredients,” its chief executive officer, Alexander Nix, told NBC News…
…It didn’t have to build everything from scratch. Mark Zuckerberg and others had already built the infrastructure the campaign needed to reach voters directly…”
“If your computer having problems or lags while using Windows operating systems, you’ll discover some fast and lightweight OS in this article. There is another post where we discussed best alternative operating systems that are not specifically for old PC’s”
By definition, the forecast of the election results is something extremely difficult. A reliable forecast does not simply consider opinion polls but it should be able to also consider the impact of historical, social and economical variables combined with various factors such as the “possible behavior” or “psychological reactions” of voters.
There is always the real risk of not considering or underestimate some essential variables that will affect the decisions of voters just on the election day.
The graphs below are based on data from sites commonly considered as reliableandtrustworthy but, in no case these charts can be regarded as scientific or reliable and are merely the result of a data processing described in a post published yesterday via Medium.
As empirically described by the Technical University of Munich through the paper “The mere number of tweets reflects voter preferences and comes close to traditional election polls”, the below analysis assumes the existence of a direct relationship between the number of tweetsgeneratedduring an electoral contest by a candidate and the final election results.
To mitigate the supposed direct relationship between the number of tweets and the final electoral results I considered other data variables as described in the post published yesterday via Medium.
“… Christian Martinez, Facebook’s head of multicultural, posted a response to the article.
“A nonprofit that’s hosting a career fair for the Hispanic community can use Facebook ads to reach people who have an interest in that community. And a merchant selling hair-care products that are designed for black women can reach people who are most likely to want its products,” he wrote.
“That merchant also may want to exclude other ethnicities for whom their hair care products are not relevant—this is a process known in the ad industry as “exclusion targeting.”…”
“…Artificial intelligence and decision-support algorithms that can offer data-driven suggestions will unleash a new level of productivity among workers, allowing everyone to focus on what matters and to continually help one another improve.
Turning this into reality may be closer than you think, thanks to machine learning and predictive data engines…”
“…But pseudocides are rarer in recent times. “Vanishing” oneself is more difficult; the world is simply too small a place now, connected as it is by social media and the surveillance it entails….”
“…Let’s say you are hiding in Japan, and a tourist takes a photo where you’re in the background,” he told me. “The photo is uploaded to social media and a week later, a cop uploads your photo into a facial recognition site like TinEye [a reverse-image search engine]. Boom—you’re busted, because TinEye will find your photo online…”