Journal 44

 June 3 - June 9

This week, I learned more about linear regression and the complete machine learning workflow involved in building and evaluating a predictive model. We went over how to prepare data for machine learning by handling missing values, creating new variables, encoding categorical variables using dummy variables, and splitting the data into training and testing sets. I also learned how to train a linear regression model, interpret the model coefficients, generate predictions, and evaluate the model using metrics such as mean squared error (MSE) and mean absolute error MAE). I really enjoyed learning about the importance of comparing a model's performance to a baseline prediction. Before this week, I understood that lower error values were ideal, but I had not thought about how to determine whether a model was actually providing useful predictions. Comparing our linear regression model to a simple baseline that predicted the average value helped demonstrate that the model was capturing meaningful relationships in the data. I also found it interesting to interpret the coefficients of the model and think about what positive and negative parameter values might imply in the context of the problem. 

This week's homework was a bit more challenging because I had to do a lot of tweaking to get my graphs to look like the sample outputs in the pdf. Also, another challenge I encountered was understanding how all of the preprocessing steps fit together. There were several pieces involved, including imputing missing values, grouping categories, creating dummy variables, and preparing NumPy arrays for Scikit-Learn. It was easy to focus on each step individually without appreciating how they collectively contribute to building a successful predictive model. Working through the notebook helped me see the overall process more clearly and reinforced the importance of careful data preparation. 

This week really strengthened my understanding of regression modeling and gave me more confidence in applying machine learning techniques to real-world datasets. These topics will be valuable in my future goal as predictive modeling is used throughout professional sports to estimate player performance, injury risk, contract value, and many other outcomes. Understanding how to prepare data, build models, evaluate their performance, and interpret their results will be essential skills in that type of role. This week's material helped me gain a better appreciation for the fact that building a machine learning model involves much more than simply fitting an algorithm, it requires thoughtful decisions at every stage of the process. Although some aspects of the workflow were challenging at first, completing the homework and interpreting the results helped connect the individual concepts into a larger picture of how predictive analytics is performed in practice.