Logit Model_Predictive Regression

[Khaladdin]

Hi all,

I have a general question related to logit regression model. I want to know your mind about one point related to logit regression model. I am using logit model for my analyzing. My key independent variable is the lagged; hence, it is predictive regression. So, my dependent variable is 0 and 1. These 0 and 1 is related to the time of the day. To make it more clear, I have minutely time-series data for one day. In one day, we have 24 hours * 60 minutes = 1440 rows as my dependent variable. The dependent variable is "0" from 1st to 600th rows and from 900th to 1440th rows - it is "normal period". However, the dependent variable is "1" from 600th to 900th rows - let's say it is "abnormal period". In its simplest from, my dependent and independent variables are:

Number of rows    Dependent variable      Independent variable
1                                   0                                         0.1
2                                   0                                         0.15
.                                    .                                           .
.                                    .                                           .
600                              0                                          0.17
601                              1                                          0.13
.                                   .                                           .
.                                   .                                           .
900                             1                                           0.22
901                             0                                           0.19
.                                  .                                            .
.                                  .                                            .
1440                           0                                           0.21

I am interested the examine the predictive power of my key independent variable. Now I use the first lag of my independent variable as my KEY VARIABLE. Actually, I am confusing about the result of my regression. The issue is, I guess that the majority part of the predictive power of lagged variable will come from the "abnormal period" rather than "normal period".
What do you think about that?

[alexchien]

the correlation between the dep var and the lagged dep var is almost 1 for your data. Therefore, I don't think it is appropriate to include the lagged dep var to the model. You will get an almost perfect fit but useless model.

thanks

alex

[Khaladdin]

I do not use lagged dependent variable. I am going to use the lagged of my independent variable, not dependent variable. Why does the correlation matter?

[Khaladdin]

Now, I got your point 🙂 Thanks