How to train a model to predict picks using multiple independent dataframes?

Question

I have a dictionary of 100+ dataframes all have the same shape [9999 rows x 4 columns] with the following columns: ['time', 'response', 'arrival_1','arrival_2']

arrival_1,arrival_2 are binary columns where only one row has a value of 1 while all the other 9998 have 0. The column basically tells us that at this time this is our arrival_1.

I am trying to build two models, one to predict 'arrival_1' and the other to predict 'arrival_2'.

However, I need the model to tell me the time where arrival_1 is 1. If I divide my data like the following:

X = ['time','response'] , y1 = ['arrival_1']

In the above case, I will need to provide the time and response to know whether y1_predict = 1

How can I go about this ? Knowing that the model needs to utilize all dataframes.