If anyone encountered a similar problem set, please point me in the right direction.

Hi, i have a prediction challenge where i'm given historical ticket sales data of a train station and i'm to use that to predict the number of tickets over the next few week.
I've used Random forest but despite all my tweaks and feature engineering i'm stuck at a rank 103 (as of today) on the leaderboard.
I used ANN model for this and the the accuracy wasnt improving to even 0.1 which was far worse than RF.
If anyone encountered a similar problem set, please point me in the right direction.