gistfile1.txt

# Rolling Average
# Moving window week
mw = df[df$Date >= "2009-11-01" & df$Date <= "2009-11-07",]

# MW days
mw_sat <- subset(mw, wday(mw$Date) == 2)
mw_mon <- subset(mw, wday(mw$Date) == 7)

# MW time
mw_sat_window <- mw_sat[mw_sat$Time >= '07:00:00' & mw_sat$Time <= "09:00:00",]
mw_mon_window <- mw_mon[mw_mon$Time >= '07:00:00' & mw_mon$Time <= "09:00:00",]

# Plot the unsmoothed data (gray)
x <- (strptime(mw_sat_window$Time, '%H:%M:%S'))
y <- mw_sat_window$Global_active_power
plot(x, y, type="l", col=grey(.5), main="Global_active_power MA from 7am-9am, 2009-11-02",sub = "(2 sided MA, 10 levels)", xlab="Time", ylab="Global_active_power",)

f20 <- rep(1/10,10)
f20

y_sym <- stats::filter(y, f20, sides=2)
lines(x, y_sym, col="blue")

legend("topleft", inset=.01,
       c("Moving Average","Original Data"), fill=c("blue","grey"), horiz=FALSE)

legend("topright", inset=.01,
       c("Minor Anomaly","Major Anomaly", "Extreme Anomaly"), fill=c("yellow","orange","red"), horiz=FALSE)


stddev <- sd(y_sym, na.rm = TRUE)
threshold_minor <- 1*stddev
threshold_major <- 2*stddev
threshold_extreme <- 4*stddev

minor_occurances <- 0
major_occurances <- 0
extreme_occurances <- 0

for(i in 1:length(y_sym)) {
  if (is.na(y_sym[i])) {
    # print("na")
  } else {
    difference <- abs(y[i] - y_sym[i])
    if (difference >= threshold_minor) {
      if (difference >= threshold_major) {
        if (difference >= threshold_extreme) {
          points(strptime(mw_sat_window[i, "Time"], '%H:%M:%S'), y[i], col = "red", pch = 19)
          extreme_occurances = extreme_occurances + 1;
          next
        }
        points(strptime(mw_sat_window[i, "Time"], '%H:%M:%S'), y[i], col = "orange", pch = 19)
        major_occurances = major_occurances + 1;
        next
      }
      points(strptime(mw_sat_window[i, "Time"], '%H:%M:%S'), y[i], col = "yellow", pch = 19)
      minor_occurances = minor_occurances + 1;
      next
    }
    next
  }
}

print(minor_occurances)
print(major_occurances)
print(extreme_occurances)

# =======

# Plot the unsmoothed data (gray)
x <- (strptime(mw_mon_window$Time, '%H:%M:%S'))
y <- mw_mon_window$Global_active_power
plot(x, y, type="l", col=grey(.5), main="Global_active_power MA from 7am-9am, 2009-11-07",sub = "(2 sided MA, 10 levels)", xlab="Time", ylab="Global_active_power",)

f20 <- rep(1/10,10)
f20

y_sym <- stats::filter(y, f20, sides=2)
lines(x, y_sym, col="blue")

legend("topleft", inset=.01,
       c("Moving Average","Original Data"), fill=c("blue","grey"), horiz=FALSE)

legend("topright", inset=.01,
       c("Minor Anomaly","Major Anomaly", "Extreme Anomaly"), fill=c("yellow","orange","red"), horiz=FALSE)


stddev <- sd(y_sym, na.rm = TRUE)
threshold_minor <- 1*stddev
threshold_major <- 2*stddev
threshold_extreme <- 4*stddev

minor_occurances <- 0
major_occurances <- 0
extreme_occurances <- 0

for(i in 1:length(y_sym)) {
  if (is.na(y_sym[i])) {
    # print("na")
  } else {
    difference <- abs(y[i] - y_sym[i])
    if (difference >= threshold_minor) {
      if (difference >= threshold_major) {
        if (difference >= threshold_extreme) {
          points(strptime(mw_sat_window[i, "Time"], '%H:%M:%S'), y[i], col = "red", pch = 19)
          extreme_occurances = extreme_occurances + 1;
          next
        }
        points(strptime(mw_sat_window[i, "Time"], '%H:%M:%S'), y[i], col = "orange", pch = 19)
        major_occurances = major_occurances + 1;
        next
      }
      points(strptime(mw_sat_window[i, "Time"], '%H:%M:%S'), y[i], col = "yellow", pch = 19)
      minor_occurances = minor_occurances + 1;
      next
    }
    next
  }
}

print(minor_occurances)
print(major_occurances)
print(extreme_occurances)