Moving forward with the course lecture, we delved into the course structure and our first project topic: Linear regression. Personally, I find it essential to revisit the fundamentals before diving deep into any subject, so I went back to the course material. Upon reading, I realized the significance of grasping the data thoroughly when employing statistical methods for analysis. Connecting with the data is vital to gain insights into its inherent nature. Since our data originates from a real source, it’s imperative that our predictions reflect a realistic approach rather than blindly fitting it into an overly simplified model.
It’s crucial to acknowledge that real-world data carries inherent errors, and these errors should be accorded due consideration to preserve the authenticity of the data. I came across Karl Gauss’s Linear Least Squares model, which calculates the absolute error values and minimizes them to approximate the data points within a linear model. Nevertheless, this model exhibits instability and unreliability. My plan is to begin by plotting individual models based on available data points, then progress to establishing a correlation between obesity and inactivity to predict diabetes percentages accurately.