So the model is y=13,81x1-0.33x2-239,8x3+5823x5 (where x5=0 or x5=1) From the co
ID: 3052625 • Letter: S
Question
So the model is y=13,81x1-0.33x2-239,8x3+5823x5 (where x5=0 or x5=1)
From the correlation matrix we get
How can we propose an alternate model?
SUMMARY OUTPUT Regression Statistics Multiple FR R Square Adjusted R Square Standard Error Observations 0.887690681 0.787994745 0.786182734 14455.19858 473 ANOVA MS Significance F Regression Residual Total 4 468 472 3.63472E+11 97789894479 4.61262E+11 90867939600 434.8731 4.2973E-156 208952766 Coefficients Standard Error P-value Lower 95% Upper 95%ower 95.09 pper 95.09 1692.076674 1796.831692.08 1796.83 12.01626312 15.59682 12.01626 15.59682 0.484283051 0.18982 -0.48428 -0.18982 300.8890541 -178.759 300.889-178.759 2763.064288 8884.472 2763.064 8884.472 t Stat 52.3768947 13.80654259 0.33705054 239.824039 5823.768042 887.7419525 0.911062475 0.074925741 31.07561975 1557.573772 0.059000135 0.952977 15.1543313 1.71E-42 -4.498461221 8.64E-06 7.717433825 7.23E-14 3.738999812 0.000208 Intercept x2 X3 x5Explanation / Answer
As per the assumptions of regression analysis, it is required that dependent variable should be linearly related to each of the independent variables. to avoid the problem of multicollinearity, it is required that none of the independent variables are strongly correlated to each other.
From the Correlation Matrix, I observed that the variables X1 X2 are strongly related to each other. this induces multicollinearity in the data.
It is recommended to apply transformations like sqrt, log, lag so as to reduce the problem of multicollinearity in the data. Thus I propose a new model as:
y=bo + b1*ln(x1) - b2*x2 - b3x3 + b5*x5
Related Questions
drjack9650@gmail.com
Navigate
Integrity-first tutoring: explanations and feedback only — we do not complete graded work. Learn more.