Question

Some statisticians prefer complex models, models that try to fit the data as closely as one can. Others prefer a simple model. They claim that although simpler models are more remote from the data yet they are easier to interpret and thus provide more insight. What do you think? Which type of model is best to use? When formulating your answer to this question you may think of a situation that involves inference that you do and need to present to other people. Would the consumers of your analysis benefit more from you having used a complex model of from yo having used a simpler model? What would be the best way to report your findings and explain them to the consumers?

Answer 1

Answer:

Here I think the inclination to intricate or basic models relies upon the point of view of the examination.

For instance in the event that we are breaking down the information for arrangement creators where we need straightforward answers like whether demise rate in North Carolina has expanded after some time, or whether more weapon laws can lessen number of crimes we should focus on basic models where explicit replies answers to these inquiries can be given.

Then again on the off chance that we are attempting to comprehend the component of an arbitrary variable and attempt to construct a prescient model I figure progressively complex model ought to get the inclination as in nature nothing is static and neither the procedures work directly.

For instance on the off chance that we need to arrange the clients dependent for them and acknowledge subtleties as fortunate or unfortunate borrower numerous highlights ought to be remembered for the model and afterward some mind boggling characterization strategy like irregular woods and so on can be utilized to get the best classifier which in since quite a while ago run will give us a hazard evaluation of the clients.

By and by there ought to be a harmony between basic model and complex model: the exchange off is called stinginess.

This is the explanation individuals in measurements presently use LASSO sort of thing where you continue adding highlights to the model and yet put a punishment with the goal that complete number of powerful highlights won't surpass a given edge.