Assignmet 1

part1/complete.R

@@ -0,0 +1,14 @@
+complete <- function(directory, id = 1:332){
+  obs <- c()
+  for (x in id){
+    a <- file.path(directory, formatC(x, width = 3, flag = "0"))
+    b <- paste(a, "csv", sep = ".")
+    data <- read.csv(b)
+#    a <- data[!is.na(data[,2]) & !is.na(data[,3]),]
+    a <- sum(complete.cases(data))
+#    obs <- c(obs, length(a[[1]]))
+    obs <- c(obs, a)
+  }
+  df = data.frame(id, obs)
+  df
+}

part1/corr.R

@@ -0,0 +1,14 @@
+corr <- function(directory, threshold = 0) {
+      matrix <- c()
+      for (x in 1:332){
+        a <- file.path(directory, formatC(x, width = 3, flag = "0"))
+        b <- paste(a, "csv", sep = ".")
+        data <- read.csv(b)
+        len <- sum(complete.cases(data))
+        if (len > threshold){
+          c <- cor(data[complete.cases(data), "sulfate"], data[complete.cases(data), "nitrate"])
+          matrix  <- c(matrix, c)
+        }
+      }
+      matrix
+}

part1/mean.R

@@ -0,0 +1,14 @@
+pollutantmean <- function(directory, pollutant, id = 1:332){
+  Sum = 0
+  length = 0
+  for (x in id){
+    a <- file.path(directory, formatC(x, width = 3, flag = "0"))
+    b <- paste(a, "csv", sep = ".")
+    data <- read.csv(b)
+    Sum = Sum + sum(data[[pollutant]], na.rm = TRUE)
+    length = length + length(data[[pollutant]][!is.na(data[[pollutant]])])
+  }
+  mean = Sum/length
+  mean
+
+}