(if this question doesn't fit here, please point me to right direction. Thanks.)
We need to check a sales data S’s accuracy to determine if it can be used in downstream calculations. Currently, it's simply checked against last year's data (if it's below/above certain %). I wanted to include history data(the data before last year) into the play. Below it's what I would like to consider(suppose we use True as accurate thus can be used):
- If it’s first time S presented without history data, we say it’s true 100%
- If S is within +/- 40% of last year’s data, we say it’s false with 80% chance the conclusion is right
- If S is within MIN to MAX range of all history data (excluding last year’s), it’s true with 85% chance of correctness
- If 2, and there are 2 too in history, then it’s false with 35% chance of being correct
Assume all history data are accurate and no other facts than history data alone is available.
I’m confused, how can I form a True/False conclusion toward S along with confidence(/chance of correctness) %?
(this is probably the whole picture of the question. No other facts - region, brand,.. available for this data.)