Context My grandma has been suffering from diabetes for over 15 years. Last summer I collected her glucose data from her meter, and hand copied all of her notes that include insulin intakes, what she ate at night and so on... The format is one day one row, since the number of measures each day is different, this format implies that there are a lot of missing values. It was later found that insulin is becoming less potent with time which might explain the last august measure. Many things to do here, I unsuccessfully tried to predict the level of early morning glucose level in order to prevent hypoglycemia. More recent data will come in a month or so if there is interest.