Regression HW
Grace Bowersox
2025-04-17
Introduction:
My goal for this regression project is to predict the sale price of residential properties.
Thesis
I believe that home type, neighborhood, sale year, and number of bedrooms will significantly influence the sale price of a home.
Data Description:
The ames data set provides 80+ variables that impact sale price of homes.
Method:
I built a multiple linear regression model to predict sale_price using the variables: home type, neighborhood, sale year, and number of bedrooms
Key Findings:
- Looking at my R-squared value, my model predicts 61% of variation in home sales prices. Since I only choose 4 factors for my model I believe this is pretty good.
- The variable I created for a single family home (one_family_house) seems to have a really great positive impact on my model, increasing price by $34,098.
- Neighborhood seems to vary greatly on the way it impacts my model, for instance the Stone Brook increases price by 104,434 whereas the neighborhood South and West of Iowa State University decreases price by 104,947.
- Sale year does not seem to have a huge impact on price, with inflation this seems to be a very normal pricing trend.
- The number of bedrooms seems to have a positive effect on sale price, I did go ahead and cap bedrooms to ensure outliers did not skew the data.
ames <- read_csv("ames.csv")## Rows: 2930 Columns: 79
## ── Column specification ────────────────────────────────────────────────────────
## Delimiter: ","
## chr (46): MSSubClass, MSZoning, Street, Alley, LotShape, LandContour, Utilit...
## dbl (33): LotFrontage, LotArea, YearBuilt, YearRemodAdd, MasVnrArea, BsmtFin...
##
## ℹ Use `spec()` to retrieve the full column specification for this data.
## ℹ Specify the column types or set `show_col_types = FALSE` to quiet this message.ames <- ames %>%
janitor::clean_names()
ames <- ames %>%
mutate(one_family_house = if_else(bldg_type == "OneFam", 1, 0))ames <- ames %>%
mutate(bedroom_capped = ifelse(bedroom_abv_gr > 6, 6, bedroom_abv_gr))
m <- lm(sale_price ~ one_family_house + year_sold + neighborhood + bedroom_capped, data = ames)predicted <- predict(m, ames)
rmse <- sqrt(mean((predicted - ames$sale_price)^2))
rsq <- summary(m)$r.squared
ggplot(ames, aes(x = bedroom_capped, y = sale_price)) +
geom_jitter() +
geom_smooth() +
labs(title = "Relationship between Sale Price and Number of Bedrooms")## `geom_smooth()` using method = 'gam' and formula = 'y ~ s(x, bs = "cs")'## Warning: Failed to fit group -1.
## Caused by error in `smooth.construct.cr.smooth.spec()`:
## ! x has insufficient unique values to support 10 knots: reduce k.