HW9 - Machine learning and Kaggle

In this assignment, we will run the machine learning prediction model, and submit the code to a kaggle competition. Using the housing prices data, we will predict the prices of homes based on characteristics.

Load and clean the data

# Import training and testing data

train_raw <- read.csv2("home-data-for-ml-course/train.csv", sep = ",",
                        stringsAsFactors = TRUE)
dim(train_raw)

## [1] 1460   81

test_raw <- read.csv2("home-data-for-ml-course/test.csv", sep = ",",
                        stringsAsFactors = TRUE)
dim(test_raw)

## [1] 1459   80

# skim(train_raw)

Note: The train dataset has one more column than the test dataset (SalePrice, which we want to predict).

Minimial cleaning

We need to replace NAs with the most frequent level or median

# Replace NAs with most frequent level or median

replace_na_most <- function(x){
    fct_explicit_na(x, na_level = names(which.max(table(x))))
}
replace_na_med <- function(x){
    x[is.na(x)] <- median(x,na.rm = TRUE)
    x
}
cleanup_minimal <- function(data){
    nomis <- data %>%
      mutate_if(is.factor, replace_na_most) %>%
      mutate_if(is.numeric, replace_na_med)
    nomis
}
train_minclean <- cleanup_minimal(train_raw)
test_minclean <- cleanup_minimal(test_raw)

Tree algorithm

Now, let us run the simplest tree algorithm and export the predictions.

mod_rpart <- rpart(SalePrice ~ ., data = train_minclean)

# Tree plot
fancyRpartPlot(mod_rpart, caption = NULL, palettes = "YlGn")

# Export the predictions
pred_rpart <- predict(mod_rpart, newdata = test_minclean)
submission_rpart <- tibble(Id = test_raw$Id, SalePrice = pred_rpart)
head(submission_rpart)

## # A tibble: 6 x 2
##      Id SalePrice
##   <int>     <dbl>
## 1  1461   118199.
## 2  1462   151246.
## 3  1463   185210.
## 4  1464   185210.
## 5  1465   249392.
## 6  1466   185210.

write_csv(submission_rpart, file = "submission_rpart.csv")

Leaderboard position

Leaderboard position