Created
March 4, 2018 01:47
-
-
Save davidjwiner/e18d2fbfc8ed86d220720ea141e3cbdc to your computer and use it in GitHub Desktop.
Team just working on my fitness's code
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
rm(list=ls()) | |
source("http://www.stanford.edu/~bayati/oit367/T367_utilities_13_alpha.R") | |
setwd("/Users/davidjwiner/Dropbox/GSB/OIT\ 367/Prediction_Challenge") | |
training = read.csv("training.csv") | |
# randomizing the training data | |
training = training[sample(nrow(training)),] | |
test = read.csv("test.csv") | |
#----------------------------------------------------------------------- | |
#----------------- Some basic preprocessing | |
# Creating a dummy response for the test set, it will never be used but helps us stack | |
# both training and test on top of each other and preprocess them together. | |
test$RESPONSE = rep(0,nrow(test)) | |
alldata = rbind(training,test) | |
# Creating a dummy for Male and one for Female. Note that we are not dropping one of these, as | |
# we did in the past, since there exist a third possibility (NA) which would be the third and | |
# dropped category | |
alldata$GenderF = as.numeric((alldata$Gender=="F")&(!is.na(alldata$Gender))) | |
alldata$GenderM = as.numeric((alldata$Gender=="M")&(!is.na(alldata$Gender))) | |
alldata$Gender=NULL | |
alldata$Profile_Status <- factor(alldata$Profile_Status) | |
alldata$City_Id <- factor(alldata$City_Id) | |
# We didn't end up using this approach | |
# threshold = 5000 | |
# factor_features = c("Source","Profile_Status", "Language", "City_Id", "Country_Code", "Region", "Sub_Region") | |
# | |
# for (col in factor_features){ | |
# levels(alldata[, col])[table(alldata[, col]) < threshold] = 'Other' | |
# } | |
#--------------------------------------------------------------------- | |
# Partition data into training and validation sets | |
train.size = 0.8 | |
train.ind = runif(nrow(training)) < train.size | |
trainingProcessed = alldata[train.ind, ] | |
validationProcessed = alldata[!train.ind, ] | |
testProcessed = alldata[-(1:nrow(training)),] | |
# Build decision tree model | |
model.tree = buildModel("RESPONSE", columns, trainingProcessed, type = 'classify', method = "decisionTree") | |
prp(model.tree, extra=4, fallen.leaves=TRUE, type=1, box.col=rainbow(30), varlen=0,digits=6,faclen=0) | |
# Output validation predictions | |
validationPredictions = predict(model, newdata = validationProcessed, type='classify') | |
pred_ROCR <- prediction(validationPredictions, validationProcessed$RESPONSE) | |
auc_ROCR <- performance(pred_ROCR, measure = "auc") | |
auc_ROCR <- auc_ROCR@y.values[[1]] | |
cat("The ROC value is ", auc_ROCR) | |
# Make test predictions and write output to file | |
predictions = genPred(model.tree, newdata=testProcessed, method='decisionTree') | |
submission = data.frame(Id=test$Id, Prediction = predictions) | |
write.csv(submission, file = "team_submission.csv", row.names = FALSE) |
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment