SpeciesPrediction_PredictiveModelling

train<-read.csv("SpeciesPresenceTrain.csv")
test<-read.csv("SpeciesPresenceTestFeatures.csv")

set.seed(2)


train_from_train <- train[1:600,2:16]
target_a_train_from_train <- train[1:600,17]
test_from_train <- train[601:800,2:16]
target_a_test_from_train <- train[601:800,17]

library(class)

knn_pred_a <-knn (train = train_from_train, test = test_from_train, cl= target_a_train_from_train, k = 29)
table(knn_pred_a, target_a_test_from_train)
mean(knn_pred_a != target_a_test_from_train)
knn_pred_a_ON_TEST <-knn (train = train[,2:16], test = test[2:16], cl= train[,17], k = 29)
knn_pred_a_ON_TEST


target_b_train_from_train <- train[1:600,18]
target_b_test_from_train <- train[601:800,18]
knn_pred_b <-knn (train = train_from_train, test = test_from_train, cl= target_b_train_from_train, k = 29)
table(knn_pred_b, target_b_test_from_train)
mean(knn_pred_b != target_b_test_from_train)
knn_pred_b_ON_TEST <-knn (train = train[,2:16], test = test[2:16], cl= train[,18], k = 29)
knn_pred_b_ON_TEST



target_c_train_from_train <- train[1:600,19]
target_c_test_from_train <- train[601:800,19]
knn_pred_c <-knn (train = train_from_train, test = test_from_train, cl= target_c_train_from_train, k = 29)
table(knn_pred_c, target_c_test_from_train)
mean(knn_pred_c != target_c_test_from_train)
knn_pred_c_ON_TEST <-knn (train = train[,2:16], test = test[2:16], cl= train[,19], k = 29)
knn_pred_c_ON_TEST


target_d_train_from_train <- train[1:600,20]
target_d_test_from_train <- train[601:800,20]
knn_pred_d <-knn (train = train_from_train, test = test_from_train, cl= target_d_train_from_train, k = 29)
table(knn_pred_d, target_d_test_from_train)
mean(knn_pred_d != target_d_test_from_train)
knn_pred_d_ON_TEST <-knn (train = train[,2:16], test = test[2:16], cl= train[,20], k = 29)
knn_pred_d_ON_TEST



target_e_train_from_train <- train[1:600,21]
target_e_test_from_train <- train[601:800,21]
knn_pred_e <-knn (train = train_from_train, test = test_from_train, cl= target_e_train_from_train, k = 29)
table(knn_pred_e, target_e_test_from_train)
mean(knn_pred_e != target_e_test_from_train)
knn_pred_e_ON_TEST <-knn (train = train[,2:16], test = test[2:16], cl= train[,21], k = 29)
knn_pred_e_ON_TEST


target_f_train_from_train <- train[1:600,22]
target_f_test_from_train <- train[601:800,22]
knn_pred_f <-knn (train = train_from_train, test = test_from_train, cl= target_f_train_from_train, k = 29)
table(knn_pred_f, target_f_test_from_train)
mean(knn_pred_f != target_f_test_from_train)
knn_pred_f_ON_TEST <-knn (train = train[,2:16], test = test[2:16], cl= train[,22], k = 29)
knn_pred_f_ON_TEST



target_g_train_from_train <- train[1:600,23]
target_g_test_from_train <- train[601:800,23]
knn_pred_g <-knn (train = train_from_train, test = test_from_train, cl= target_g_train_from_train, k = 29)
table(knn_pred_g, target_g_test_from_train)
mean(knn_pred_g != target_g_test_from_train)
knn_pred_g_ON_TEST <-knn (train = train[,2:16], test = test[2:16], cl= train[,23], k = 29)
knn_pred_g_ON_TEST



target_h_train_from_train <- train[1:600,24]
target_h_test_from_train <- train[601:800,24]
knn_pred_h <-knn (train = train_from_train, test = test_from_train, cl= target_h_train_from_train, k = 29)
table(knn_pred_h, target_h_test_from_train)
mean(knn_pred_h != target_h_test_from_train)
knn_pred_h_ON_TEST <-knn (train = train[,2:16], test = test[2:16], cl= train[,24], k = 29)
knn_pred_h_ON_TEST



target_i_train_from_train <- train[1:600,25]
target_i_test_from_train <- train[601:800,25]
knn_pred_i <-knn (train = train_from_train, test = test_from_train, cl= target_i_train_from_train, k = 29)
table(knn_pred_i, target_i_test_from_train)
mean(knn_pred_i != target_i_test_from_train)
knn_pred_i_ON_TEST <-knn (train = train[,2:16], test = test[2:16], cl= train[,25], k = 29)
knn_pred_i_ON_TEST



target_j_train_from_train <- train[1:600,26]
target_j_test_from_train <- train[601:800,26]
knn_pred_j <-knn (train = train_from_train, test = test_from_train, cl= target_j_train_from_train, k = 29)
table(knn_pred_j, target_j_test_from_train)
mean(knn_pred_j != target_j_test_from_train)
knn_pred_j_ON_TEST <-knn (train = train[,2:16], test = test[2:16], cl= train[,26], k = 29)
knn_pred_j_ON_TEST


target_k_train_from_train <- train[1:600,27]
target_k_test_from_train <- train[601:800,27]
knn_pred_k <-knn (train = train_from_train, test = test_from_train, cl= target_k_train_from_train, k = 29)
table(knn_pred_k, target_k_test_from_train)
mean(knn_pred_k != target_k_test_from_train)
knn_pred_k_ON_TEST <-knn (train = train[,2:16], test = test[2:16], cl= train[,27], k = 29)
knn_pred_k_ON_TEST


target_l_train_from_train <- train[1:600,28]
target_l_test_from_train <- train[601:800,28]
knn_pred_l <-knn (train = train_from_train, test = test_from_train, cl= target_l_train_from_train, k = 29)
table(knn_pred_l, target_l_test_from_train)
mean(knn_pred_l != target_l_test_from_train)
knn_pred_l_ON_TEST <-knn (train = train[,2:16], test = test[2:16], cl= train[,28], k = 29)
knn_pred_l_ON_TEST


target_m_train_from_train <- train[1:600,29]
target_m_test_from_train <- train[601:800,29]
knn_pred_m <-knn (train = train_from_train, test = test_from_train, cl= target_m_train_from_train, k = 29)
table(knn_pred_m, target_m_test_from_train)
mean(knn_pred_m != target_m_test_from_train)
knn_pred_m_ON_TEST <-knn (train = train[,2:16], test = test[2:16], cl= train[,29], k = 29)
knn_pred_m_ON_TEST

#################################################################################

install.packages("mldr")
library(mldr)

train_mldr <-mldr_from_dataframe(train[601:800,],labelIndices = c(17,18,19,20,21,22,23,24,25,26,27,28,29))
head(train_mldr)

#Don't Run MldrGUI until required
mldrGUI()


predictions <- c (as.integer(knn_pred_a)-1 ,as.integer(knn_pred_b)-1, as.integer(knn_pred_c)-1,
                  as.integer(knn_pred_d)-1, as.integer(knn_pred_e)-1, as.integer(knn_pred_f)-1,
                  as.integer(knn_pred_g)-1, as.integer(knn_pred_h)-1, as.integer(knn_pred_i)-1,
                  as.integer(knn_pred_j)-1, as.integer(knn_pred_k)-1, as.integer(knn_pred_l)-1,
                  as.integer(knn_pred_m)-1)
dim(predictions)

predictions_matrix <- matrix(predictions, nrow = 200, ncol=13)
colnames(predictions_matrix)<-c("a","b","c","d","e","f","g","h","i","j","k","l","m")
head(predictions_matrix)

dim(predictions_matrix)
summary(train_mldr)

res <- mldr_evaluate(train_mldr, predictions_matrix)
plot(res$ROC, main = "ROC curve for species")
res$AUC
res$ROC

test_predicted <- c (test$id,
                     as.integer(knn_pred_a_ON_TEST)-1 ,as.integer(knn_pred_b_ON_TEST)-1, as.integer(knn_pred_c_ON_TEST)-1,
                     as.integer(knn_pred_d_ON_TEST)-1, as.integer(knn_pred_e_ON_TEST)-1, as.integer(knn_pred_f_ON_TEST)-1,
                     as.integer(knn_pred_g_ON_TEST)-1, as.integer(knn_pred_h_ON_TEST)-1, as.integer(knn_pred_i_ON_TEST)-1,
                     as.integer(knn_pred_j_ON_TEST)-1, as.integer(knn_pred_k_ON_TEST)-1, as.integer(knn_pred_l_ON_TEST)-1,
                     as.integer(knn_pred_m_ON_TEST)-1)

test_predicted_matrix <- matrix(test_predicted, nrow = 260, ncol=14)
colnames(test_predicted_matrix)<-c("id","a","b","c","d","e","f","g","h","i","j","k","l","m")
head(test_predicted_matrix)
tail(test_predicted_matrix)

write.csv(test_predicted_matrix, file = "KNN_test_predicted.csv",row.names=FALSE)