# ZAD1
auto <- read.csv("http://ls.home.amu.edu.pl/data_sets/Automobile.csv", 
                 sep = ",", header = TRUE, na.strings = "?")
head(auto)
auto$num.of.doors <- ifelse(auto$num.of.doors == "four", 4, 2)

auto_wna <- na.omit(auto)
cat("wymiar nowych danych")
dim(auto_wna)

auto_wna_sel <- subset(auto_wna, select = c(horsepower, city.mpg, peak.rpm,
                                            curb.weight, num.of.doors, price))
pairs(auto_wna_sel)
model_1 <- lm(price ~ horsepower + city.mpg + peak.rpm + curb.weight +
                num.of.doors, data = auto_wna)
model_1

coef(model_1)
confint(model_1)

summary(model_1)

fitted(model_1)
residuals(model_1)

step(model_1)
step(model_1, k = log(nrow(auto_wna)))
model_0 <- lm(price ~ 1, data = auto_wna)
step(model_0, direction = "forward", scope = formula(model_1))
step(model_0, direction = "forward", scope = formula(model_1), k = log(nrow(auto_wna)))

model_1_1 <- lm(price ~ horsepower + city.mpg + curb.weight + num.of.doors, data = auto_wna)
summary(model_1_1)$coefficients
summary(model_1_1)$adj.r.squared
model_1_2 <- lm(price ~ horsepower + curb.weight + num.of.doors, data = auto_wna)
summary(model_1_2)$coefficients
summary(model_1_2)$adj.r.squared
model_1_3 <- lm(price ~ horsepower + curb.weight, data = auto_wna)
summary(model_1_3)$coefficients
summary(model_1_3)$adj.r.squared

auto_sel <- subset(auto, select = c(horsepower, city.mpg, peak.rpm,
                                    curb.weight, num.of.doors, price))
summary(auto_sel)
install.packages("Hmisc")
library(Hmisc)
auto_sel$price <- as.numeric(impute(auto_sel$price, mean))
auto_sel$horsepower <- as.numeric(impute(auto_sel$horsepower, mean))
auto_sel$peak.rpm <- as.numeric(impute(auto_sel$peak.rpm, mean))
auto_sel$num.of.doors <- as.numeric(impute(auto_sel$num.of.doors, median))
summary(auto_sel)



cat("2.", "\n")
pairs(auto_sel)
model_1_i <- lm(price ~ horsepower + city.mpg + peak.rpm + curb.weight +
                  num.of.doors, data = auto_sel)
model_1_i
coef(model_1_i)
confint(model_1_i)
summary(model_1_i)
fitted(model_1_i)
residuals(model_1_i)

cat("3.", "\n")
step(model_1_i)
step(model_1_i, k = log(nrow(auto_sel)))
model_0_i <- lm(price ~ 1, data = auto_sel)
step(model_0_i, direction = "forward", scope = formula(model_1_i))
step(model_0_i, direction = "forward", scope = formula(model_1_i), k = log(nrow(auto_sel)))

cat("4.", "\n")
model_1_i_1 <- lm(price ~ horsepower + city.mpg + curb.weight + peak.rpm, data = auto_sel)
summary(model_1_i_1)$coefficients
summary(model_1_i_1)$adj.r.squared
model_1_i_2 <- lm(price ~ horsepower + city.mpg + curb.weight, data = auto_sel)
summary(model_1_i_2)$coefficients
summary(model_1_i_2)$adj.r.squared
model_1_i_3 <- lm(price ~ horsepower + curb.weight, data = auto_sel)
summary(model_1_i_3)$coefficients
summary(model_1_i_3)$adj.r.squared

new_data <- data.frame(curb.weight = 2823, horsepower = 154)
model_2 <- lm(price ~ curb.weight + horsepower, data = auto_wna)
model_2_i <- lm(price ~ curb.weight + horsepower, data = auto_sel)
stats::predict(model_2, new_data, interval = "prediction")
stats::predict(model_2_i, new_data, interval = "prediction")
summary(model_2)$adj.r.squared
summary(model_2_i)$adj.r.squared