[GroupID] Group 11 Boston Housing price

Groups

• 曾偉綱 資科碩二 108753122
• 盧禹叡 經濟碩二 109258026
• 張修誠 資科碩一 110753165
• 邱顯安 資科碩一 110753110

Goal

A breif introduction about your project, i.e., what is your goal?

Demo

Commend to reproduce our result

Rscript performance.R --fold <k> --train data/training --test data/test --report results/performance.csv --predict result/predict.csv

• Shiny io app : https://brianchiu.shinyapps.io/finalproject/

Folder organization and its related information

docs

• Your presentation, 1101_datascience_FP_<yourID|groupName>.ppt/pptx/pdf, by Jan. 13
  • 
• Any related document for the final project
  • papers
  • software user guide

data

• Form Kaggle API : $ kaggle competitions download boston-housing
• Input format : CSV
• Preprocessing
  • Check Missing value

  colSums(is.na(data))

  No missing value
  • Outlier check & remove by box plot
  • Skewness check & process
  • Correlated Heat Map between Features

code

• Which method do you use?
  • We use Random forest model for our prediction.

  train_control <- trainControl(method = "none")
  
  model <- train(medv~., data = train_data, method = "rf", trControl = train_control)

• What is a null model for comparison?
  • We compare our model with different select respectively

model <- train(medv~., data = train_data, method = "knn", trControl = train_control)
model <- train(medv~., data = train_data, method = "lm", trControl = train_control)

• How do your perform evaluation? ie. cross-validation, or addtional indepedent data set
  • We use cross-validation to evaluate our performance, and also use the addtional indepedent dataset to check our prediction on Kaggle.

results

• Which metric do you use
  • We use RMSE as our metric.
• Is your improvement significant?
  • Yes, we improve the RMSE from 3.45 to 3.41
• What is the challenge part of your project?
  • Test chose the most appropriate features

References

• Packages you use

  library(caret)
  library(randomForest)
  library(ggvis)
  library(shiny)

• Related publications
  • Kaggle
    • https://www.kaggle.com/c/boston-housing/overview
  • shiny templete
    • https://shiny.rstudio.com/gallery/
  • package
    • https://www.rdocumentation.org/packages/randomForest/versions/4.6-14/topics/importance