4.2 Prediction using Regression

Caution: Be careful extrapolating results beyond the range of the dataset

Prediction Interval (Uncertainty around a single value)

• Confidence Intervals (Uncertainty around a statistic)

• Individual data point error

Here is an example of individual data point error. If we filter for those properties that have four bedrooms, 3 bathrooms and a lot square footage between 10k and 11k, the sale price varies by $50k. This is error in our model.

dat %>% dplyr::filter(Bedroom_AbvGr == 4 & 
                 bath == 3 & 
                 Lot_Area >=10000 & 
                 Lot_Area < 11000) %>% 
  dplyr::select(Sale_Price) %>%
  dplyr::arrange(-Sale_Price)

## # A tibble: 9 × 1
##   Sale_Price
##        <int>
## 1     218500
## 2     211000
## 3     170000
## 4     165150
## 5     157000
## 6     139000
## 7     127500
## 8     103500
## 9     100000