tengpeng

users.gender.replace('-unknown-', np.nan, inplace=True)

tengpeng

users_nan = (users.isnull().sum() / users.shape[0]) * 100
users_nan[users_nan > 0].drop('country_destination')

tengpeng

users[users.age > 122]['age'].describe()

tengpeng

users.loc[users.age > 95, 'age'] = np.nan

tengpeng

## Convert categorical features into numerical features ##
for (f in common_vars) {
  if (class(train[[f]]) == "character"){
    levels <- unique(c(train[[f]], test[[f]]))
    train[[f]] <- as.integer(factor(train[[f]], levels=levels))
    test[[f]]  <- as.integer(factor(test[[f]],  levels=levels))
  }
}

tengpeng

nci = subset(nci, select = c("CNS", "RENAL", "BREAST", "NSCLC", "MELANOMA", "OVARIAN", "LEUKEMIA", "COLON"))
nci = nci[, c("CNS", "RENAL", "BREAST", "NSCLC", "MELANOMA", "OVARIAN", "LEUKEMIA", "COLON")]
nci = nci[, which(colnames(nci) %in% c("CNS", "RENAL", "BREAST", "NSCLC", "MELANOMA", "OVARIAN", "LEUKEMIA", "COLON"))]

tengpeng

result = if(any(result <= 1.5)) 1 else 2
result = ifelse(result <= 1.5, 1, 2)

tengpeng

prediction <- as.numeric(pred > 0.5)

tengpeng

err <- mean(as.numeric(pred > 0.5) != test$label)

tengpeng

library(xgboost)

X = as.matrix(train[,-58])
y = train[,58]
numberOfClasses <- max(y) + 1

bst <- xgboost(data = X, label = y, max.depth = 2, eta = 1, nthread = 2, nround = 200, objective = "binary:logistic", verbose = 1)

pred = predict(bst, data.matrix(test[,-58]))
pred = as.numeric(pred > 0.5)