Timeline for What is a good model for this data?(plot attached)
Current License: CC BY-SA 4.0
12 events
| when toggle format | what | by | license | comment | |
|---|---|---|---|---|---|
| Jun 9, 2020 at 19:24 | comment | added | amarykya_ishtmella | @Rodolphe my data would. be something like an inverse gaussian dist. | |
| Jun 9, 2020 at 18:09 | comment | added | Rodolphe | Now that you have clarified a bit your assumptions / mental model, it is obviously not a linear relationship that your are looking for. Your model should include in some way the fact that you are hypothesising a plateau. If you do not know what the equation of your model should look like, you should search the relevant expert litterature about it. Maybe if that doesn't help try to construct it through first principles of your study-object i.e. process-based modeling and optimization. | |
| Jun 9, 2020 at 18:02 | comment | added | Rodolphe | Thanks for clarifying. So when you talk about plateau you mean that period during which the dependent variable is at its highest before decreasing again ? Or the period after decreasing when it is back and stable at a higher than beginning value ? | |
| Jun 9, 2020 at 11:56 | comment | added | amarykya_ishtmella | @Rodolphe yes...The dependent variable can't reduce to zero. It increases over trials and comes back to a level above its usual activity level.Example, when a child is growing it reached maximum height during adulthood and as this person grows old, the height might reduce to that of being an adulthood, but it doesn't reduce back to that of a baby | |
| Jun 8, 2020 at 22:41 | comment | added | Rodolphe | Why do you make the plateau-hypothesis ? Is there some theory backing it up ? Because your data here doesn't look like it is going to plateau so this hypothesis must surely come from a preceding theory ? | |
| Jun 8, 2020 at 6:27 | comment | added | amarykya_ishtmella | @Rodolphe Plateau here is like a tabletop(.that there is no variance over time) | |
| Jun 3, 2020 at 0:32 | comment | added | Rodolphe | Thanks for your answers. Maybe try to think differently. Could you state formally the definition of what is a plateau ? Then how to see if your data becomes more like your definition, towards the last trials than at the beginning (maybe using a sliding window of trials) ? | |
| Jun 2, 2020 at 22:08 | comment | added | amarykya_ishtmella | @Rodolphe I'm asking wrt to whether should I use a linear regression for this hypothesis testing? | |
| Jun 2, 2020 at 22:05 | comment | added | amarykya_ishtmella | @HarveyMotulsky in both conditions it's 30 people each. This what you see is my plotted data. My hypothesis is that the data over trails converge to a tabletop/plateau in both conditions. How do I test this hypothesis for data converging to a tabletop.Plus the plotted data doesn't look like a linear model. | |
| Jun 2, 2020 at 20:14 | comment | added | Harvey Motulsky | If you want suggestions for a model that matches data, you need to show data (and explain what the data are). How large a sample size does your graph show? | |
| Jun 2, 2020 at 19:49 | comment | added | Rodolphe | You talk about a model that assumes a plateau. However you ask for a good model ? Please clarify your question / problem. What do you actually want to know ? | |
| Jun 2, 2020 at 19:07 | history | asked | amarykya_ishtmella | CC BY-SA 4.0 |