-1
$\begingroup$

I am relative new in regression analysis. I would like to know, if there is a way in regression analyis to estimate the risk or calculate the risk for future values?

An example:

We want to predict the export of a country for next year. We can use linear regression to estimate the export value for the next year. This value may be influenced by another parameter for example Weather catastrophe. Is there any way to estimate such a risk in linear regrssion?

Improve this question
$\endgroup$
3
  • $\begingroup$ Do you have data that includes weather catastrophes? $\endgroup$
    – Ian_Fin
    Commented Jul 21, 2016 at 12:29
  • $\begingroup$ Actualy not, it is only an assumption $\endgroup$
    – Kaja
    Commented Jul 21, 2016 at 12:42
  • 1
    $\begingroup$ regression is based on using the data at hand to infer the relationship between variables and consequently generate predictions. If you have no data at hand (about weather catastrophes) you can't infer a relationship nor make any predictions $\endgroup$
    – Ian_Fin
    Commented Jul 21, 2016 at 12:45

1 Answer 1

Reset to default
0
$\begingroup$

You can try a multiple regression, using weather catastrophe. I am not sure how you are measuring "catastrophe" but assuming it is binary (0 or 1), you can use this parameter as a categorical predictor. In running the regression analysis you can then tell whether or not the weather catastrophe is a significant predictor. If it is, then in your analysis you can compare predictions for the export given a catastrophe and without a catastrophe. I believe this is the risk in which you were hoping to measure in your forecasts.

forecast <- lm(export ~ predictor + weather, data)
summary(forecast)
new1 <- data.frame(predictor=--, weather=0)
new2 <- data.frame(predictor=--, weather=1)
predict(forecast, data=new1)
predict(forecast, data=new2)
Improve this answer
$\endgroup$
3
  • $\begingroup$ Very nice explanation, To the point you have described $\endgroup$
    – Kaja
    Commented Jul 21, 2016 at 12:45
  • 1
    $\begingroup$ @Kaja this describes how to use data about weather catastrophes to make predictions about values. In your comment to the question you say that you do not have data about weather catastrophes. How can this be said to be best answer to your question? $\endgroup$
    – Ian_Fin
    Commented Jul 21, 2016 at 12:48
  • 1
    $\begingroup$ @Lan_Fin as I said I have no data for weather catastrphe. There is a weather catastroohe or not?(1 or 0) and I would like to find the impact of this important parameter on my prediction. I found the answer very nice. What I was looking for. I must test it in detail :-) $\endgroup$
    – Kaja
    Commented Jul 21, 2016 at 13:05

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.