Discussion about this post

User's avatar
Michael Cook's avatar

In the "Time Trap" note you took two sets of random data and then added a constant to all data. This makes it non-random data and there is true correlation. For example you have 2 production lines each with 10 people making ginger cookies using 20 different dough cutters. These are randomly packed into boxes. The weight of the boxes will be random but trends over time will be linear with random fluctuations i.e. your first example today. But suppose the dough mixer water supply starts getting blocke up. The thickness of the cookies will decrease and the boxes will gradually get lighter. This will show random variation but with linear trends downwards for both production lines. The time sequence correlation between the two production lines is a true correlation pointing to an unknown variable. The cookie factory manager needs to recognise and fix this before he looses all the contracts. Michael

Expand full comment
3 more comments...

No posts