Good afternoon,
I am completed a linear regression model which I personally am happy with the
output and residual charts (although they are a bit bunched). I have been
using visreg() to visualise my data, and although I have logged the dependent
variable (as it is cost data), on one of the graphical outputs I am getting
negative fitted values for one of the Continents. Attached is the data used,
below is the code for the model and graph. If anyone has any ideas I would
hugely appreciate it, I tried a glm with log link too and this didn't help.
LM22 <- lm(logTotal ~ logArea + Continent + factor(Method1) + Popdens,
data=dummy)
Output Residuals: Min 1Q Median 3Q Max -1.22742 -0.31675
-0.00909 0.28885 1.20224
Coefficients: Estimate Std. Error t value Pr(>|t|)
(Intercept) -1.40972 0.77661 -1.815 0.081506 . logArea
0.03811 0.04041 0.943 0.354619 ContinentAsia 3.51165
0.80009 4.389 0.000182 ***ContinentAustralasia 3.48549 0.97433 3.577
0.001454 ** ContinentEurope 2.44829 0.82369 2.972 0.006452 **
ContinentGlobal 2.37910 1.00668 2.363 0.026197 * ContinentNorth
America 2.27953 0.62960 3.621 0.001303 ** ContinentSouth America 3.60997
0.77627 4.650 9.22e-05 ***factor(Method1)2 1.11259 0.38386
2.898 0.007696 ** factor(Method1)3 0.82519 0.47470 1.738 0.094459 .
factor(Method1)4 -0.06737 0.67341 -0.100 0.921104 Popdens
-0.14261 0.31675 -0.450 0.656435 ---Signif. codes: 0 �***� 0.001
�**� 0.01 �*� 0.05 �.� 0.1 � � 1
Residual standard error: 0.6151 on 25 degrees of freedom (12 observations
deleted due to missingness)Multiple R-squared: 0.6998, Adjusted
R-squared: 0.5677 F-statistic: 5.297 on 11 and 25 DF, p-value: 0.0002714
visreg(LM22) gives me a graph with "Africa" being within the realm of negative
"Total" which is impossible. If you tell me how can provide the data and
graphs, but the email was kicked back with them in.
[[alternative HTML version deleted]]
______________________________________________
[email protected] mailing list
https://stat.ethz.ch/mailman/listinfo/r-help
PLEASE do read the posting guide http://www.R-project.org/posting-guide.html
and provide commented, minimal, self-contained, reproducible code.