Learning Algorithm

Every step has an initial weight of 1 and the initial learning rate is also 1. Every time a user finishes a step, we will record the actual time it took for the user. According to the weight, we will update the estimated time for that step. We increase the weight if the real time differs from the estimated time. Every time you learn, learning rate will decrease by 1⁄4. If actual time differs more than 2X than the estimated time. The maximum estimated time change will be 2X. This is so that we can decrease outlier such as if a person forgot to click next step and let it sit there or clicked too fast accidentally.