This morning, the four-year U.S. election is in full swing again.
Although it’s irrelevant, as a qualified melon-eater, many people must be refreshing the results of the U.S. election from time to time. After all, guessing that the president of Country M is much more interesting than writing jokes.
On the one hand, there is a person called Jianguo, who claims that no one understands Trump better than himself, and on the other is Biden, the sleepy king who has been a vice president for several years and is suspected of Alzheimer’s. These two people are running together. The President of the United States is dramatic enough, and what makes this election even more interesting is the “face-slap” and “reverse” that are never late.
Big data and polls
In fact, long before the start of this election, casinos, polls, economists, analysts, big data companies, etc. have already begun to predict, and this year’s scenario seems to be quite similar to the Trump fight against Hillary four years ago. :
All the media, polling agencies and big data forecast results show that the 2016 election will undoubtedly be won by Hillary Clinton, who will become the first female president in American history.
However, the final result was surprising. At first, Trump, who was not optimistic at all, won 6 swing states in one breath, fought back in a desperate situation, and successfully staged a bloody counter-kill!
And as a result, the big data system and polls were immediately scolded by dog-blood sprayers, and the world’s melon-eating people said they were very deceived.
It coincides with this year that the United States has experienced several chaotic events such as the black movement, the new crown epidemic, and the immigration crisis.
As a result, Trump’s approval rate has been declining. The famous American media The Economist even predicted that Biden’s win rate in the general election will exceed 97%.
The big data system and the polling agency also stated that this year’s forecasting algorithm has been revised, and the forecast result will be more than 80% reliable, and the last oblong event will not occur.
So before this year’s general election, what are the prediction results given by the polling big data system?
Biden will win the general election and be elected as the new President of the United States. Trump will encounter Waterloo
The basis for polling big data seems to be very scientific. Various polling agencies will regularly conduct surveys on people in different regions. However, it combines the rules of the US general election, the regional division of the Democratic and Republican parties, based on the general election rules of previous years, and combines big data. Forecasting algorithm, finally get a more reasonable result
Looking at it this way, it seems that the results of polling big data will be very reasonable
Slap face as always
In fact, it seems to be the case, because before the general election, Biden was basically certain that he had 210 votes in hand, which is only a small margin from the US Constitution’s “270 votes for president.”
Trump’s situation is much more miserable. As long as one of the six swing states does not get it, it is very likely to lose the election.
If Trump wants to win, he must win all the swing states from Texas to Pennsylvania. The difficulty is unimaginable.
Among them, the most important key ticket warehouses-Oklahoma, Texas, Pennsylvania, Florida, Biden’s polling situation is dominant, it can be said that Biden almost has a great advantage.
On the other hand, Biden only needs to win Pennsylvania to seal the victory. The general election seems to have been fixed in advance.
However, no one expected that today’s US election would be so thrilling and dramatic.
The election has just begun. Biden’s competition with Trump is indeed very fierce. Biden’s advantage is very strong, and many swing states have begun to appear light blue, indicating that Biden’s support rate is relatively high.
Even the key ticket warehouse in Texas has turned blue. I guess no one will doubt the accuracy of the polls.
However, only one morning passed, and the masses of melon-eating people discovered that those swing states that turned blue turned light red, and finally turned red!
Trump once again won a complete victory in the swing state. As of the time of posting, Trump has basically won the key votes in the swing state.
Although Biden still has an advantage of 12 votes, anyone with a discerning eye can see that there are very few votes that Biden can get in the back of the voting warehouse.
The one who wins the swing state wins, this is an unchanging truth under the US election rules
This “farce”, in all likelihood, Trump ended up as a winner.
Even if there are some irresistible incidents that allow Biden to win by a narrow margin, this election is enough to make the U.S. polling big data system lose its face!
Is big data useless?
Two consecutive mistakes made the credibility of the polls’ big data quickly fall to the point of heinously, and even turned into a rat crossing the street, everyone shouted and beaten.
It stands to reason that the big data system of polls in the United States should be quite scientific, but why are there so many oblong incidents?
In fact, many people will have a misunderstanding that the authenticity of the results of a big data system comes from the high degree of fit and fit between the algorithm and the actual situation.
But when we are doing data analysis, data mining or other data-related work, we must first ensure the authenticity and reliability of the data source
Take this U.S. election as an example. The data source that the polling big data system relies on is the polling situation of each state, as well as the state’s data in previous years, such as which states have always been the Democratic Party’s votes and which The state has always been a hardcore Republican
But the authenticity of this data source is very worthy of investigation:
For example, is it possible to support Biden verbally but vote for Trump in secret?
Of course there is, and there may be many such cases, because everyone knows how bad Trump’s radical supporters are.
For example, will there be survivor errors?
Of course, a large part of Trump’s supporters are farmers and ranchers, and they are mostly distributed in remote mountainous areas. The big data system cannot guarantee the effective proportion of this population in the polls.
For example, will the historical data of each state produce errors?
Of course, it will, especially in recent years, the immigration situation between the states in the United States has been quite frequent. Take Texas as an example. Texas has always been the home of the Republican Party. However, this year’s competition for votes in Texas has been extremely fierce.
The reason is that many people can’t stand the high tax environment in California. The wealthy and homeless pour into Texas at the same time, causing the rapid expansion of Democratic voters. This is the deviation of historical data.
Of course, there are many other situations that can cause data to be biased. They all lie in the processing of the data source. Whether it is really convincing is the most difficult and important task in data analysis.
These tasks are far from being solved by those polling big data agencies changing the weighting coefficients, but making this prediction even more outrageous.
Therefore, the US election this time not only slapped the media in the face, but also severely slapped those polling big data agencies that are sitting and waiting to die. The cart is upside down, the trustworthiness is upside down, and it is self-defeating!