Comment Varied Dropdown with airSlate SignNow

Comment varied dropdown

[Music] in this video we're going to talk about model building and variable selection and before i get into a bit of a guide to doing it i want to start with some just general comments so the first thing i want to point out is that there is no right way of doing it there are wrong ways of doing it but there's not one right way of doing it if there was the right way to build your model and select variables we wouldn't really need to think much about it we would have built an algorithm to do it plug in our data return a model final model to us so that's what these automative automated model variable selection procedures are trying to achieve and there are weaknesses in doing those so let me just kind of write this down there is no right way but there are wrong ways and one one thing i'd like to say to try and draw this point home if you gave two of the top researchers in their field the exact same set of data the exact same research question they're quite likely going to come up with different models and different variables in there and slightly different approaches to trying to answer that question right so again i just do want to stress this there's not one right way of doing it there definitely are wrong things to do but there's not one correct way to do it the approach i like to think of it as being half science half art so again there's a combination of using statistics and numerical approaches and there's a combination of using thinking and creativity in there as well one thing you want to incorporate is using prior knowledge so use what you already know about this subject area or this research question you're looking at if you know certain variables are confounders make sure you measure those when you collect data and adjust for them in your model incorporate the things you already know one important one the approach depends on your goal and what i mean by that is if you're building an effect size model you're trying to estimate the effect of some variable on the outcome or if you're trying to build a predictive model if you're trying to predict the outcome so we're going to go through and first look at some approaches or some guides to building a model second variable when you want to estimate the effect of x1 on some outcome and then we're also going to look briefly at a guide or some suggestions on approaches to selecting variables and building a model when the goal is to predict them one that i cannot stress enough and i really want to stress it because in a course especially one like this where we're in a classroom um well i guess we're not in the classroom right now we're kind of in a virtual...