Timeline for Explaining gam model output with a reference
Current License: CC BY-SA 4.0
10 events
| when toggle format | what | by | license | comment | |
|---|---|---|---|---|---|
| Feb 19, 2023 at 15:52 | comment | added | Ahsk | Thank you very much. | |
| Feb 19, 2023 at 15:16 | comment | added | EdM | @Ahsk that doesn't seem to describe the modeled values on the plot on the right, although it might describe the raw values on the left. Modeled values at 12 degrees are about the same as they are at 17. If you are talking about the modeled values, I also wouldn't describe them as "observed." For the modeled values you might say something like: "With other variables taken into account, the model suggests a dip in the response as a function of temperature between about 14 and 16 degrees." | |
| Feb 19, 2023 at 11:53 | comment | added | Ahsk |
In addition to providing the reference, maybe I should reword my interpretation too to avoid potential conflict. Can I say that, Overall, increasing temperatures had a positive effect on the response variable, although the effect was not linear. Lower values for response variable were observed between 14-16 degree Celsius. Thanks very much.
|
|
| Feb 18, 2023 at 1:33 | comment | added | Ahsk | Okay, thank you. | |
| Feb 16, 2023 at 14:35 | comment | added | EdM |
@Ahsk I'm looking for a way to generate something like a "raw data" plot that looks more like your smooth. The raw data plot you show doesn't take into account any of the other variables in the model. If you could take into account the contributions of mean_rh and avg_daily_rain, that might help. Try plotting the residuals from a simple linear model, with just those two linear predictors, against the mean_temp values as an alternative to your simple raw data plot on the left. I suspect that will look more like the smooth on the right.
|
|
| Feb 16, 2023 at 1:08 | comment | added | Ahsk |
I have already added residuals using plot(mod1, rug = TRUE, residuals = TRUE, pch = 1, cex = 1, shade = TRUE, seWithMean = TRUE, shift = coef(mod1)[1]). Unless you're referring to some other residuals?
|
|
| Feb 16, 2023 at 1:07 | vote | accept | Ahsk | ||
| Feb 15, 2023 at 20:12 | history | edited | EdM | CC BY-SA 4.0 |
added 582 characters in body
|
| S Feb 15, 2023 at 20:07 | history | answered | EdM | CC BY-SA 4.0 | |
| S Feb 15, 2023 at 20:07 | history | made wiki | Post Made Community Wiki by EdM |