# set up file name
# look at data in text editor
source_data = 'house_prices_in_19083.txt'
homes = read.csv2(source_data, stringsAsFactors = FALSE, dec='.')
# review structure of homes data set
str(homes)
# look for weird values / distributions
summary(homes)
# can edit directly if needed
fix(homes)
# better to edit in code
homes<-homes[!(homes$sqft==1),]
# clean out rows with missing data
homes = na.omit(homes)
# build and assess model
lm.allpredictors = lm(price ~ ., data = homes)
summary(lm.allpredictors)
lm.resid = resid(lm.allpredictors)
lm.resid
# are residuals distributed as expected?
plot(lm.resid)
plot(density(resid(lm.allpredictors)))
# build a model just based on square feet
# plot price vs sqft
plot(homes$sqft, homes$price)
lm.sqft = lm(price ~ sqft, data = homes)
summary(lm.sqft)
abline(lm.sqft)
# which point is which?
identify(homes$sqft, homes$price, paste(homes$price, ' ', homes$sqft))
# predict how much a home will cost
predict(lm.allpredictors, data.frame(beds=c(4), baths=c(3), sqft=c(3694), lotsz=c(16553)),
        interval="prediction")
predict(lm.allpredictors, data.frame(beds=c(3), baths=c(1.5), sqft=c(1499), lotsz=c(5271)),
        interval="prediction")