From 23436db628c5e494ab985ef877319e4c96995565 Mon Sep 17 00:00:00 2001
From: ibishof <ibishof@gmail.com>
Date: Thu, 11 Jan 2024 11:46:56 -0800
Subject: [PATCH] Delete Machine_Learning/svm.R

---
 Machine_Learning/svm.R | 99 ------------------------------------------
 1 file changed, 99 deletions(-)
 delete mode 100644 Machine_Learning/svm.R

diff --git a/Machine_Learning/svm.R b/Machine_Learning/svm.R
deleted file mode 100644
index ea6e624..0000000
--- a/Machine_Learning/svm.R
+++ /dev/null
@@ -1,99 +0,0 @@
-library(e1071)
-library(matrixStats)
-library(BBmisc)
-library(psycho)
-library(dplyr)
-library(tidyverse)
-help(package="e1071")
-
-
-setwd("C:\\Users\\kqjf682\\Omics_pipelines\\CKD\\module_reduced\\module_reduced_CKD_Diabetes\\svm")
-train <- read.csv("training.csv")
-train <- train[,-1]
-validation <- read.csv("test.csv")
-validation <-validation[,-1]
-
-# Normalize
-#
-train <- normalize(train, method = "range", range = c(0, 1), margin = 1L, on.constant = "quiet")
-validation <- normalize(validation, method = "range", range = c(0, 1), margin = 1L, on.constant = "quiet")
-
-# Scale data by calculating z-score
-# 
-trainN <- train[2:ncol(train)]
-train_mean <- colMeans(trainN)
-train_mean <- sweep(trainN, 2, train_mean, FUN = '-')
-trainsd <- t(as.matrix(colSds(as.matrix(trainN))))
-trainz <- sweep(train_mean, 2, trainsd, FUN = '/')
-train <- cbind(train$Diagnosis, data.frame(trainz))
-
-validationN <- validation[2:ncol(validation)]
-validation_mean <- colMeans(validationN)
-validation_mean <- sweep(validationN, 2, validation_mean, FUN = '-')
-validationsd <- t(as.matrix(colSds(as.matrix(validationN))))
-validationz <- sweep(validation_mean, 2, validationsd, FUN = '/')
-validation <- cbind(validation$Diagnosis, data.frame(validationz))
-
- train <- train %>% 
-  rename(
-    Diagnosis = `train$Diagnosis`
-      )
- 
- validation <- validation %>% 
-   rename(
-     Diagnosis = `validation$Diagnosis`
-   )
- 
-
-
-# Build the actual model
-# The smaller gamma is, the more the hyperplane is going to look like a straight line. If gamma is too great, the hyperplane will be more curvy 
-# cost = the penaility of constraints violation
-# eps-regression is default for regression models, nu-regression can be selected also
-# (default) C-classification, nu-classification, one-classification (for novelty detection)
-# Kernal defualt is radial; linear, polynomial, and sigmoid are also options
-model <- svm(Diagnosis ~ ., data = train, gamma = 1e-05, cost = .01, type = "C-classification", kernel = "linear")
-summary(model)
-
-# Test model on training regression
-pred <- predict(model, train)
-cor(pred,soma_dt$GFR_0, method = "pearson")
-table(pred, y)
-
-
-# Test model for training classification
-pred <- predict(model, train)
-train_pred <- as.factor(pred)
-tyyyyy <- ordered(train_pred)
-roc_ojb_train <- roc(response =  train$Diagnosis, predictor =tyyyyy )
-roc_ojb_train$auc
-
-# Test model on validation
-#validation <- read.csv("test.csv")
-pred <- predict(model, validation)
-cor(pred,validation$GFR_0, method = "pearson")
-table(pred, y)
-
-#Get AUC
-pred <- predict(model, validation)
-zzzz <- as.factor(pred)
-yyyyy <- ordered(zzzz)
-roc_ojb_MS <- roc(response =  validation$Diagnosis, predictor =yyyyy )
-roc_ojb_MS$auc
-
-write.csv(pred, "svm_MS.csv")
-write.csv(answers, "answers-svm_ms.csv")
-
-
-# Tune svm parameters
-tuned_parameters <- tune.svm(Diagnosis~., data = train, kernel = "linear", gamma = 10^(-5:-1), cost = 10^(-3:1))
-summary <-summary(tuned_parameters )
-summary
-
-# COde to gain feature importance (Needs work)
-library(rminer)
-M <- fit(y~., data=soma_dt, model, kpar=list(sigma=0.10), C=2)
-svm.imp <- Importance(M, data=soma_dt)
-
-
-