In: Statistics and Probability
11.The number of meals consumed in a school canteen is recorded each day for two weeks.
week | day | number of meals |
1 | monday | 252 |
tuesday | 265 | |
wednesday | 281 | |
thursday | 242 | |
friday | 229 | |
2 | monday | 258 |
tuesday | 270 | |
wednesday | 289 | |
thursday | 251 | |
friday | 237 |
n-point moving average value for n is 5.
(i) a)Explain why a suitable value for n is 5
ii) b) explain why centring will not be necessary
iii) c) Calculate all the 5-point moving average values and insert them in appropriate places in the table above.
d)Use values from your table to find an estimate for the seasonal component for Wednesday
i). a). Since, we have 5 days a week considered. Moving averages are an estimate of trend pattern that remove seasonality. Here, 5 days a week will show a seasonal pattern over week. Taking average of 5 observations will eliminate the possible seasonal pattern over week and will be able to give true estimate of trend.
ii). b. As, n = 5 is an odd number it will correspond to 3rd observation (that wll be centered itself) or centering will not be necessary.
iii). c).
week | day | number of meals | 5 MA |
1 | monday | 252 | |
tuesday | 265 | ||
wednesday | 281 | 253.8 | |
thursday | 242 | 255 | |
friday | 229 | 256 | |
2 | monday | 258 | 257.6 |
tuesday | 270 | 259.4 | |
wednesday | 289 | 261 | |
thursday | 251 | ||
friday | 237 |
is the 5 point moving average corresponding to 3rd - 8th observation.
d). We can take the moving averages to estimate the seasonality.
week | day | number of meals | 5 MA | y - 5MA | y / 5MA |
1 | monday | 252 | |||
tuesday | 265 | ||||
wednesday | 281 | 253.8 | 27.2 | 1.11 | |
thursday | 242 | 255 | -13 | 0.95 | |
friday | 229 | 256 | -27 | 0.89 | |
2 | monday | 258 | 257.6 | 0.4 | 1.00 |
tuesday | 270 | 259.4 | 10.6 | 1.04 | |
wednesday | 289 | 261 | 28 | 1.11 | |
thursday | 251 | ||||
friday | 237 |
There are two types of model: Additive and Multiplicative.
Addtive ; y = trend + seasonality + error
Here, seasoanlity is estimated by column y - 5MA (actual - 5MA)
Multiplicative: y = trend*seasonality*error
Here, seasonality is estimated by column y/5MA = (actual/5MA)
Please rate my answer and comment for doubt.