R Notebook

hist(sample$gas_price)

summary(sample$gas_price)

    Min.  1st Qu.   Median     Mean  3rd Qu.     Max. 
    0.00    41.00    71.39   102.31   125.00 34912.03 

under_one = sample$gas_price >= 1
sample2 <- sample[under_one, ]

summary(sample2$gas_price)

   Min. 1st Qu.  Median    Mean 3rd Qu.    Max. 
    1.0    41.0    71.5   102.3   125.0 34912.0 

third <- sample2$gas_price <= 125
third <- sample2[third,]
hist(third$gas_price)

fourth <- sample2$gas_price > 125
fourth <- sample2[fourth, ]
hist(fourth$gas_price)

fourth2 <- sample2$gas_price <= 241
fourth2 <- sample2[fourth2,]
hist(fourth2$gas_price)

fourth3 <- sample2$gas_price > 241
fourth3 <- sample2[fourth3, ]
summary(fourth3$gas_price)

   Min. 1st Qu.  Median    Mean 3rd Qu.    Max. 
  241.0   264.0   305.6   425.7   415.0 34912.0 

fourth4 <- sample2$gas_price <= 415
fourth4 <- sample2[fourth4,]
hist(fourth4$gas_price)

over <- sample2$gas_price > 415
under <- length(sample2$gas_price)
over <- sum(over)

over

[1] 3095

under

[1] 199941

over/under

[1] 0.01547957

(1 - sum(under_one)/length(sample$gas_price)) + (over/under)

[1] 0.01577457

over2 <- sample$gas_price <= 415
final_sample <-sample[over2, ]

under2 <- final_sample$gas_price >= 1
final_sample <- final_sample[under2,]

summary(final_sample$gas_price)

   Min. 1st Qu.  Median    Mean 3rd Qu.    Max. 
   1.00   40.00   70.00   91.14  121.00  415.00 

hist(final_sample$gas_price)