Continuing on the previous thesis day's plan, I studied how the different types of statistical tests work. One website also included different types of software that is best used for predictive analytics. Contrary to what I said in the previous thesis blog post, I think I will use the R programming language for my research product. I already started learning the basic syntax and methods in R. Not only is it efficient for statistical analysis, it also visualizes data with few methods.
I'll admit, today was a slow day for my thesis research. I began by researching how predictive analysis is used. In "A Predictive Analytics Primer," the author writes about assumptions used in building models for trends. However, although these assumptions are useful in the short term, they can become invalid as time goes on. He mentions that time and the "lack of a key variable" are what can make assumptions no longer true. I can probably use that in my research project.
I am still narrowing down a topic. I did make some progress in choosing one, and I even have a product idea in mind. I would like to focus predictive analytics on trends and data found in religion. My product would be a model predicting future trends on different religions. So far, I have looked at how megachurches use statistical analysis to evangelize and convert more people to their ideology. I can predict the church membership of megachurches and smaller churches in other denominations. I know that at the end of the year, not only do I want to predict an outcome, I want to explain the context and underlying influences that could contribute to it.
Another topic I could do would be how religious screening and banning affects the religious landscape in America. For example, because members of terrorist groups such as ISIS and Al-Qaeda are Muslim, many paranoid Americans have the conception that all Muslims are terrorists and violent. As a result, Muslims are being barred from entering the country. Supporters of the religious ban use previous studies to "prove" that by barring Muslims, they are preventing possible terrorists from entering the country. If one were to look at the bigger picture of terrorism and religion, one could see that that statement is not accurate. It is misleading. In the bigger picture, "religious identity fails to expose malicious intent." But researchers collected all this data and studied it intensely. How could they be wrong? The problem is that in statistics, especially in predictions, almost always has error. There is margin of error and bias in collecting and analyzing data. It can approximate, but it can't be entirely accurate.
I think at the moment, I'm leaning more towards the latter topic idea. I want to study how this misconception about Islam led to biased statistics and how that would change the religious landscape over the next 50 years. I'll stick to the 50-years prediction for now.
Sources Used:
https://www.predictiveanalyticstoday.com/statistical-analysis/
https://hbr.org/2014/09/a-predictive-analytics-primer
http://hackingchristianity.net/2012/03/the-temptation-of-church-analytics-who-can-resist.html
http://www.predictiveanalyticsworld.com/patimes/data-science-screening-religion-blunt-instrument-security/8818/
I'll admit, today was a slow day for my thesis research. I began by researching how predictive analysis is used. In "A Predictive Analytics Primer," the author writes about assumptions used in building models for trends. However, although these assumptions are useful in the short term, they can become invalid as time goes on. He mentions that time and the "lack of a key variable" are what can make assumptions no longer true. I can probably use that in my research project.
I am still narrowing down a topic. I did make some progress in choosing one, and I even have a product idea in mind. I would like to focus predictive analytics on trends and data found in religion. My product would be a model predicting future trends on different religions. So far, I have looked at how megachurches use statistical analysis to evangelize and convert more people to their ideology. I can predict the church membership of megachurches and smaller churches in other denominations. I know that at the end of the year, not only do I want to predict an outcome, I want to explain the context and underlying influences that could contribute to it.
Another topic I could do would be how religious screening and banning affects the religious landscape in America. For example, because members of terrorist groups such as ISIS and Al-Qaeda are Muslim, many paranoid Americans have the conception that all Muslims are terrorists and violent. As a result, Muslims are being barred from entering the country. Supporters of the religious ban use previous studies to "prove" that by barring Muslims, they are preventing possible terrorists from entering the country. If one were to look at the bigger picture of terrorism and religion, one could see that that statement is not accurate. It is misleading. In the bigger picture, "religious identity fails to expose malicious intent." But researchers collected all this data and studied it intensely. How could they be wrong? The problem is that in statistics, especially in predictions, almost always has error. There is margin of error and bias in collecting and analyzing data. It can approximate, but it can't be entirely accurate.
I think at the moment, I'm leaning more towards the latter topic idea. I want to study how this misconception about Islam led to biased statistics and how that would change the religious landscape over the next 50 years. I'll stick to the 50-years prediction for now.
Sources Used:
https://www.predictiveanalyticstoday.com/statistical-analysis/
https://hbr.org/2014/09/a-predictive-analytics-primer
http://hackingchristianity.net/2012/03/the-temptation-of-church-analytics-who-can-resist.html
http://www.predictiveanalyticsworld.com/patimes/data-science-screening-religion-blunt-instrument-security/8818/
Comments
Post a Comment