This week I mainly focus on my Insights through Data’s final group project work. Firstly, we find a dataset related to vote rate. We try to find the correlation between GDP and vote rate. However, one this is that different countries will have different time to hold the election, which is difficult for us to compare the different in the same year. Besides, we also want to compare the differences among countries. Thus, we need more data to support us to do the data analysis while some data is hard to find through a reliable source. Finally, we decided to change to another dataset named WHO. We pay attention to the relationships among variables like status (Nominal), life_expectancy (Ratio), percentage_expenditure (Ratio) and so on. This dataset is big and reliable, so we directly use this one to do the analyses. The first step is to clean the data, which means we need to remove or delete the missing data and outliers. We need to consider the following parts:
1. Detect the outliers
• Boxplots/histograms
• Tukey’s Method
2. Deal with outliers
• Drop outliers?
• Limit/Winsorize outliers?
• Transform the data using log/inverse/square root/etc?

Since each variable has a unique amount of outliers and also has outliers on different sides of the data, the best route to take is probably winsorizing (limiting) the values for each variable on its own until no outliers remain. After this step, we can do the data exploration.

Through group work, I figure out a lot of problems. I try to use code and programming to analyze realistic problem.

For another course——Clonality of Data, we use a new form to present our ideas. We choose poster which is totally a new method to solve the problem. Muslin is a absolutely new area knowledge for me. I struggled for a period. Thanks to my team member Emily, she gave me a lot of explanations and advice of the topic. I start to find it is not so hard to understand.

I changed my project’s outline. I think it’s hard to find inequality in social media and data analysis. One of the more interesting arguments I found in the course inside through data is about data bias and algorithmic bias. I think I would prefer to look at algorithmic bias. I want to find a specific example. In-depth analysis of this example will help me solve the algorithm bias research. I want to start from its causes, combined with social factors and technical dilemmas to elaborate on the algorithm bias. And then some discussion about his influence. I am currently concerned about. Gender bias in the workplace and racial inequality in the health care system. This example is quite interesting, but I still need some time to make clear which example to refer to for analysis

Week 7-8 / Knowledge Integration and Project Planning: Data, Inequality and Society (2022-2023) by blogadmin is licensed under a Creative Commons Attribution CC BY 3.0