+ 1

# Predictor variable vs. Target variable

I’m currently at the Exloratory Data Analysis of Data modelling and I’m having a hard time understanding the difference between predictor and target variables. I have already searched numerous sites but I can’t understand it. Can someone please explain it simply and (uf you may) give examples ? I can not continue with EDA if I can not differentiate the two variables.

6 Respostas

+ 3

Hi! If you want to estimate something with a modell you built, you needs some data to put in to the modell. The input data may be some kind of ovservations, that will help you to predict a modell value. So you put the data in to the modell and you get a result (a response). The data you put in to the modell are your predictor variables and the the result from the modell are your target variables.
(Input: pred.vars) -> [Model] -> (Output: targ.vars)

+ 1

The death, to me, seems to be a response of a set of factors, for example socio economical factors. Depending of how old you are, how rich or well educated you are, will the probability to die in Covid differ. So the the death (rate) may be a target variable.

+ 1

Identifying the predictors and target, is mainly about understanding the problem domain.
The end goal of this exercise, is always to make some meaningful prediction. This may be expressed as a yes/no or a percentage of likelihood, or as a set of discrete values. You want to predict the unknown, based on data that you have.
Some examples:
- a bank wants to know if it's client would be willing to invest in stocks, based on their account balance history
- a webshop wants to know which product the customer is most likely to buy, based on their purchase history
- a hospital wants to know the expected success rate of a surgical procedure, based on the patient's medical record
Can you see the pattern? :)

0

I’m doing a data modeling about covid with (City, Cases, Infected Local, Infected Forign, Date and Death as my variables) and I’m planning to use the Death variable as the predictor variable however I’m still confused on to what target variable to use. What do you think?

0

Oh that’s actually a good insight. I think the city variable is a better predictor variable? Since it is a location and can be observed

0

Or the month perhaps.