Added question 2 + export plot to image

Project/main.R

@@ -112,7 +112,7 @@ x_min <- min(min(imdb$year, na.rm = TRUE), min(groupLens$year, na.rm = TRUE), mi
 x_max <- max(max(imdb$year, na.rm = TRUE), max(groupLens$year, na.rm = TRUE), max(netflix$year, na.rm = TRUE))
 # Define colors
 
-color <- c("blue", "red", "green")
+color <- rainbow(3)
 
 imdb_year_avg <- imdb %>%
                 group_by(year) %>%
@@ -135,7 +135,7 @@ yearList <- merge(imdb_year_avg, groupLens_year_avg, by = "year")
 yearList <- merge(yearList, netflix_year_avg, by = "year")
 yearList <- mutate(yearList, mean = ((rating + rating.x + rating.y) / 3))
 
-png(filename=returnPath("output/1.png"), height = 400, width = 900, bg = "white")
+png(filename=returnPath("output/Q1.png"), height = 400, width = 900, bg = "white")
 
 plot(yearList$rating.x,
      type = "l",
@@ -146,7 +146,8 @@ plot(yearList$rating.x,
      xlab = "Years",
      ylab = "Avg. rating",
      cex.lab=0.8,
-     lwd=2
+     lwd=2,
+     main = "In what movie release year where the average ratings the highest?"
      )
 axis(1, at=1:length(yearList$year), labels = yearList$year, pos = 0)
 axis(2, las = 1, at = 2*0:y, pos = 1)
@@ -172,15 +173,40 @@ lines(yearList$rating,
 #       col = "yellow",
 #       lwd = 2
 # )
-
-legend(1, 10, set_names, cex = 0.8, col = color, lty=1:3, lwd = 2, bty="n")
+sorted <- arrange(yearList, desc(mean))
+highest <- sorted[1,]
+highest$mean <- round(highest$mean, digits = 3)
+
+points(
+  sum(
+    between(sorted$year, min(sorted$year),highest$year)
+    ),
+  highest$mean,
+  pch = 8,
+  lwd = 2,
+  cex = 2
+  )
+text(
+  sum(
+    between(sorted$year, min(sorted$year),highest$year)
+  ) - 0.3,
+  highest$mean + 0.7,
+  labels = highest$mean
+)
+text(
+  50,
+  10,
+  labels = paste(
+    "The highest average rating was in:",
+    highest$year
+  )
+)
+legend(1, 3, set_names, cex = 0.8, col = color, lty=1:3, lwd = 2, bty="n")
 
 dev.off();
 
 print("In what year are the ratings the highest?")
 
-sorted <- arrange(yearList, desc(mean))
-highest <- sorted[1,]
 print(paste("That was in:", highest$year, "Score:", highest$mean))
 
 # Cleanup
@@ -189,5 +215,36 @@ remove(highest)
 
 print("Working on question no. 2...")
 
+netflix_ <- round(mean(yearList$rating), digits = 3)
+imdb_ <- round(mean(yearList$rating.x), digits = 3)
+groupLens_ <- round(mean(yearList$rating.y), digits = 3)
+
+vct <- c(imdb_, groupLens_, netflix_)
 
+png(filename=returnPath("output/Q2.png"), height = 500, width = 450, bg = "white")
+
+barplot(
+  vct,
+  col = color,
+  names.arg = set_names,
+  ylim = c(0, y),
+  axes = TRUE,
+  xlab = "Data Provider",
+  ylab = "Avg. score",
+  main = "Which provider has the highest avrerage score?"
+  )
+
+text((1 - 0.3), (imdb_ + 0.2), labels = imdb_, col = color[1])
+text((2 - 0.1), (groupLens_ + 0.2), labels = groupLens_, col = color[2])
+text((3 + 0.1), (netflix_ + 0.2), labels = netflix_, col = color[3])
+
+text(1.5, y - 1, labels = paste("Provider with hightest average score is:", set_names[which.max(vct)]))
+
+dev.off()
+
+# Cleanup
 
+remove(netflix_)
+remove(groupLens_)
+remove(imdb_)
+remove(vct)