I have developed a database of storm events that contain a number of attributes for each storm event, e.g. the total storm rainfall depth plus the antecedent total depth of rainfall for a number of different days before the event (i.e. 60 days, 30 days, 7 days, 3 days and 1 day before). I also have a response variable being the water level (stage) at a water level gauge. When I plot just total rainfall versus stage response its all over the map (no real correlation) because I believe that the antecedent rainfall has a big effect on the water level stage reading (i.e if it rains a lot when its dry, there isnt much runoff so the water level gauge doesn't go up) but a smaller rain when its been wet has a big impact on water level reading and hence flooding.
I am not sure how to analyze this data that has multiple attributes, i.e. each data point contains values for several parameters (depths of rainfall at several different discrete times) versus a single response variable. I would like to develop equations that give the best fit of total storm rainfall and antecedent that best explains the water level at the downstream end of the watershed. In other words, if I could have a prediction of the upcoming storm, what combination of antecedent rainfall best predicts flooding.
I am hoping someone might suggest an approach and the related commands that would help me analyze the data. I have about 400 storm events and they are all in Excel so I would have to import form Excel into Mathematica and then analyze. I hope this isn't too general a question but I am not sure how to get started. Thank you for any thoughts, Roger