Thanks for the elucidating video. I have a question about the fourth assumption. If I have cross-sectional data, what could be an alternative to the Durbin-Watson test ?
hi, for binary dependent variable, you need to use logistic regression (see ua-cam.com/video/i8tjLQUPc8Y/v-deo.html). In logistic regression majority of the linear regression assumptions does not apply. The three main assumptions that apply are: (1) no multicolinearity among independent variables (see ua-cam.com/video/KB0Xxe6n8Hw/v-deo.html), (2) there are no extreme outliers (you can use boxplot, see ua-cam.com/video/LUf6Y4Wakaw/v-deo.html), and finally (3) the sample size should be large enough (200+ recommended).
Thank you, you are great! Fast and very well explained! Thank you! You concluded that you have homoscedasticity in your data, but what do you do then? What do we do if we have this problem? What kind of analysis can we run?
Homoscedasticity is good, that what we what and in that case we can run regression as usual. However, if we have heteroskedasticity, we can check for outliers, remove them or report regression results with and without outliers and compare how results differ. we can also rum robust regression estimators that minimizes the effects of heteroskedasticity.
You are FANTASTIC! You go fairly quickly but explain EVERYTHING that matters so all can be understood. EXCELLENT!
Thank you.
How do I test the linearity for a single dependent variable and 4 independent variables?
so touching for an excellent video
hi, why is the "significance of parameters" only an optional assumption for linear regression?
Thanks for the elucidating video. I have a question about the fourth assumption. If I have cross-sectional data, what could be an alternative to the Durbin-Watson test ?
Durbin-Watson test is not relevant for cross-sectional data
hi, what if the dependent variable is binary?
hi, for binary dependent variable, you need to use logistic regression (see ua-cam.com/video/i8tjLQUPc8Y/v-deo.html). In logistic regression majority of the linear regression assumptions does not apply. The three main assumptions that apply are: (1) no multicolinearity among independent variables (see ua-cam.com/video/KB0Xxe6n8Hw/v-deo.html), (2) there are no extreme outliers (you can use boxplot, see ua-cam.com/video/LUf6Y4Wakaw/v-deo.html), and finally (3) the sample size should be large enough (200+ recommended).
@@RESEARCHHUB that is so helpful.. Tq so much for your reply
Thank you, you are great! Fast and very well explained! Thank you! You concluded that you have homoscedasticity in your data, but what do you do then? What do we do if we have this problem? What kind of analysis can we run?
Homoscedasticity is good, that what we what and in that case we can run regression as usual. However, if we have heteroskedasticity, we can check for outliers, remove them or report regression results with and without outliers and compare how results differ. we can also rum robust regression estimators that minimizes the effects of heteroskedasticity.
i love you!
if sum of residual is not zero than what to do?
If you estimate a linear regression, it must be zero.
@@RESEARCHHUB For me it was not zero. Does that mean that I cannot use linear regression? What do you propose to do next?