The courts and politicians have deemed anybody stating there is fraud is a crackpot and gets banned from sharing information.
Putting the visualizations into a data platform (I am using Power BI) gets the attention of normie data scientists much better than commentary text.
Getting the message out won't change the election, but winning the minds of smart people may motivate people to push for change.
It takes away the impression that the media is creating we are idiots.
B) Hasn't all of the analysis already done and dismissed?
There are dozens of correlation areas that have not been explored due to the small time frame: For example, just a cursory query shows that the ratio of mail-in voters to in-person voters is significantly higher for areas with the same demographic in the key metropolitan areas that had an impact versus those that did not.
The evidence is overwhelmingly not for widespread fraud, but for strategic fraud which is where a lot of the prior research failed. Focusing on just the areas where Biden had huge turnouts and democrats had control of voting process have inexplicable correlations to turnout, registration, spikes, numbers of dead and ineligible voters voting far beyond statistical norms for other areas with identical demographic and polling data (I have complete address information and death information that links to those areas).
Those particular data points can be highlighted through visualizations that show the statistical likelihood using data science techniques including confidence intervals. This type of analysis is not disputable.
Using blind feature analysis and data science tools, unknown correlation (unsupervised learning) can be leveraged to find even more salient evidence than what has been presented to date.
Data visualizations that show correlations which can be viewed by data scientist to get the message out.
Time is of the essence go aggregate this data. Data sources are being removed as we speak. Pre-election polling data which significantly shows voter preferences and expected margins for Biden/Trump in different states to establish correlation studies are being purged. Ultimately my goal is to produce an academic research paper on this. For example, the fact that Houston has very similar polling as Detroit yet starkly different results from the base polling is a statistical anomaly. As you compound the number of anomalies and display in a visualization, this becomes more compelling.
Please DM me if you want to be involved with this. Per my comment, I don't think this is a waste of time, there are still reasons this work has value. Thank you.
Fuck, you are talking way above my paygrade, pede. Fully appreciate what you are doing, though and wish I could help. I'll pass the word along and see if there are folks out there who wouldn't mind pitching in.
MODS pls STICKY
dang that Social Security has 88M rows
Questions: A) Why this is still important?
The courts and politicians have deemed anybody stating there is fraud is a crackpot and gets banned from sharing information.
Putting the visualizations into a data platform (I am using Power BI) gets the attention of normie data scientists much better than commentary text.
Getting the message out won't change the election, but winning the minds of smart people may motivate people to push for change.
It takes away the impression that the media is creating we are idiots.
B) Hasn't all of the analysis already done and dismissed?
There are dozens of correlation areas that have not been explored due to the small time frame: For example, just a cursory query shows that the ratio of mail-in voters to in-person voters is significantly higher for areas with the same demographic in the key metropolitan areas that had an impact versus those that did not.
The evidence is overwhelmingly not for widespread fraud, but for strategic fraud which is where a lot of the prior research failed. Focusing on just the areas where Biden had huge turnouts and democrats had control of voting process have inexplicable correlations to turnout, registration, spikes, numbers of dead and ineligible voters voting far beyond statistical norms for other areas with identical demographic and polling data (I have complete address information and death information that links to those areas).
Those particular data points can be highlighted through visualizations that show the statistical likelihood using data science techniques including confidence intervals. This type of analysis is not disputable.
Using blind feature analysis and data science tools, unknown correlation (unsupervised learning) can be leveraged to find even more salient evidence than what has been presented to date.
Data visualizations that show correlations which can be viewed by data scientist to get the message out.
Time is of the essence go aggregate this data. Data sources are being removed as we speak. Pre-election polling data which significantly shows voter preferences and expected margins for Biden/Trump in different states to establish correlation studies are being purged. Ultimately my goal is to produce an academic research paper on this. For example, the fact that Houston has very similar polling as Detroit yet starkly different results from the base polling is a statistical anomaly. As you compound the number of anomalies and display in a visualization, this becomes more compelling.
Please DM me if you want to be involved with this. Per my comment, I don't think this is a waste of time, there are still reasons this work has value. Thank you.
Fuck, you are talking way above my paygrade, pede. Fully appreciate what you are doing, though and wish I could help. I'll pass the word along and see if there are folks out there who wouldn't mind pitching in.
Data science professionals Need to do a better job of communicating the existing evidence.