Dear Michael
Why are you looking for the sum of absolutes differences, for evaluating how similar are the data of the two weeks?
You may use several alternatives:
1. Correlation of Pearson
2. Correlation of Spearman.
3. Simple Linear Regresion.
In 1 for example you are normalizing the data during the calculations,
Your set of data are in similar range so no need for normalizing them for comparing them. Just do Pearson Correlation, because it's interval scale. and also calculate Simple Linear Regression.
But for normalizing you need mean and standard deviation of 2 set of data separately.
Then in any set, minus every Sleep Hours from mean and divide to standard deviation. you have z score.
You can multiple Z in 10 and plus with 50. It's McCall T normalization.
Normalising can mean different things to different people. Some people would use it to describe a z score transformation and others scaling on a common scale such as 0 to 1.
The answer depends on what you are trying to do and what effects you are trying to remove from the similarity comparison. If you want to remove the effect of the week (for example) you might normalise each week separately (thus prevent different in weeks from influencing the comparison).
- Badges
- Science topic
- Similar topics
- Mathematics
- Statistics
- Applied Statistics