The raw data behind the story 'Forecasting the race for the House' https://projects.fivethirtyeight.com/2018-midterm-election-forecast/house/

house_district_forecast

Format

The dataset is a data frame with 302,859 rows representing district-level results of the classic, lite, and deluxe house forecasts since 2018/08/01 and 11 variables.

forecastdate

date of the forecast

state

state of the forecast

district

district of the forecast

candidate

name of the candidate

party

party of the candidate

incumbent

whether the candidate is incumbent

model

the model of the forecast

win_probability

the probability of the corresponding party winning

voteshare

the voteshare of the corresponding party

p10_voteshare

the top 10 percentile of the voteshare

p90_voteshare

the top 00 percentile of the voteshare

Source

FiveThirtyEight’s House, Senate And Governor Models Methodology: https://fivethirtyeight.com/methodology/how-fivethirtyeights-house-and-senate-models-work/

Note

the original dataset included an empty column "special", which was removed.

See also