Key Concepts Learned
- Part 1: Where We Are
- Review
- Weeks 1-3: Data foundations
- Census data, tidycensus, spatial data basics
- Visualization and exploratory analysis
- Week 5: Linear regression fundamentals
- Y = f(X) + ε framework
- Train/test splits, cross-validation
- Checking assumptions
- Week 6: Expanding the toolkit
- Categorical variables and interactions
- Spatial features (buffers, kNN, distance)
- Neighborhood fixed effects
- The Regression Workflow
- Building the model:
- Visualize relationships
- Engineer features
- Fit the model
- Evaluate performance (RMSE, R²)
- Check assumptions
- Spatial diagnostics:
- Are errors random or clustered?
- Do we predict better in some areas?
- Is there remaining spatial structure?
- If errors cluster spatially, it suggests:
- Missing spatial variables
- Misspecified relationships
- Non-stationarity (relationships vary across space) 非平稳性
- Part 2: Understanding Spatial Patterns in Errors
- Visualizing Error Patterns（误差有聚集，不随机，不好）→怎么改进：fixed
- Part 3: Moran’s I
- Moran’s I measures spatial autocorrelation
- Range: -1(Perfect negative correlation (dispersion)) to +1(Perfect positive correlation (clustering)),0 = Random spatial pattern
- wij = spatial weight between locations i and j (0 or 1)
Coding Techniques
- [New R functions or approaches]
- [Quarto features learned]
Questions & Challenges
- What I didn’t fully understand
- the basic workflow: pull-commit-push
- Areas needing more practice
- remember the essential dplyr functions
Connections to Policy
- [How this week’s content applies to real policy work]
Reflection
- [What was most interesting]
- [How I’ll apply this knowledge]