How does the calculator work?

Every 100 blocks, we fit a Poisson regression model that estimates the expected number of blocks it will take for a transaction to confirm based on the gas price and the amount of gas used by a transaction based on data from the last 10,000 blocks. Since the relationships bewteen time and both gas price and gas used are non-linear, the gas price and the gas used are modeled as categorical variables.

For the gas price, the following categories are used in the model:
1) below average
2) average
3) above average
4) fastest

For gas used, the following categories are used:
1) median or below
2) median to 75th percentile
3) 75th to 90th percentile
4) 90th percentile and above