Timeline for Can simple linear regression be done without using plots and linear algebra?
Current License: CC BY-SA 3.0
39 events
| when toggle format | what | by | license | comment | |
|---|---|---|---|---|---|
| Mar 13, 2022 at 14:34 | answer | added | hachiko | timeline score: 0 | |
| May 28, 2016 at 20:10 | comment | added | Kyle Strand | As a software engineer who majored in math, you're absolutely right: conceptually, the visual elements in all mathematical fields are always (or almost always) merely "optional gimmicks", as you say. This is something, ideally, all mathematicians and math students, sited or not, should learn! I'm sorry your professors were not more helpful in this regard. (By the way, I came here from your "Tools of a Blind Programmer" blog post, which I loved. Thank you for writing it!) | |
| Apr 22, 2016 at 11:14 | answer | added | ctd2015 | timeline score: 1 | |
| Apr 5, 2016 at 21:11 | answer | added | Chris K | timeline score: 1 | |
| Apr 4, 2016 at 5:42 | answer | added | Glen_b | timeline score: 4 | |
| Apr 4, 2016 at 4:39 | comment | added | Parham Doustdar |
It's not compatible, in that I don't get told that the 2 is a superscript, subscript, or something else, and I have to guess. However, in StackExchange I solve that by clicking on edit and viewing the markdown code. But, typing things out makes it much easier.
|
|
| Apr 3, 2016 at 19:28 | answer | added | Vectornaut | timeline score: 2 | |
| Apr 3, 2016 at 18:43 | comment | added | Vectornaut |
A question about formatting: are LaTeX formulas like $(x_2, y_2)$ compatible with the screen-reading method you use? Do you prefer formulas to be typed out, like (x_2, y_2)?
|
|
| Apr 2, 2016 at 7:22 | vote | accept | Parham Doustdar | ||
| Apr 2, 2016 at 6:55 | answer | added | Chris Rackauckas | timeline score: 19 | |
| Apr 2, 2016 at 6:25 | comment | added | Parham Doustdar | Thank you everyone. All the answers had a roll in helping me understand what this is all about. I wish I could accept them all. However, the least I can do is to upvote them all. Thank you. | |
| Apr 2, 2016 at 6:24 | vote | accept | Parham Doustdar | ||
| Apr 2, 2016 at 7:22 | |||||
| Apr 2, 2016 at 5:07 | answer | added | Chris J | timeline score: 2 | |
| Apr 2, 2016 at 2:20 | answer | added | Diego | timeline score: 0 | |
| Apr 2, 2016 at 1:33 | comment | added | Aksakal | more I think about this question more I realize how little we know about blind people's approaches to learning math and statistics. I didn't even think how they would read the HTML page with LaTeX formulae. | |
| Apr 2, 2016 at 0:21 | history | tweeted | twitter.com/StackStats/status/716057874854776832 | ||
| Apr 1, 2016 at 22:07 | answer | added | butte | timeline score: 3 | |
| Apr 1, 2016 at 15:55 | comment | added | Aksakal | @J.M. so we need to find ways to check for sensibility without plots. there's too much looking at the plots. | |
| Apr 1, 2016 at 15:08 | comment | added | J. M. is not a statistician | "not dependent on graphs and plots." - you might still need to plot things to check for sensibility; look up Anscombe's quartet, for instance. | |
| Apr 1, 2016 at 14:58 | answer | added | Tim | timeline score: 3 | |
| Apr 1, 2016 at 14:52 | answer | added | EdM | timeline score: 10 | |
| Apr 1, 2016 at 14:41 | answer | added | Giuseppe Biondi-Zoccai | timeline score: 7 | |
| Apr 1, 2016 at 14:29 | comment | added | conjectures | Gradient descent would work. However, OLS gives you an efficient 1-iteration algorithm for finding the best unbiased MSE coefficients. | |
| Apr 1, 2016 at 14:27 | comment | added | conjectures | Also, the challenge is not really to understand the model, but to fit the model to the data (i.e. to find that 2333 is the best coefficient). | |
| Apr 1, 2016 at 14:24 | comment | added | conjectures | You don't need to use a plot. Indeed, for multiple linear regression (regression with many predictors) you can't plot a $p+1$ dimensional space. However, the linear algebra still works. All linear algebra formulae involved in linear regression can be reduced to operations on simple scalar numbers. You just wouldn't want to do it that way by hand if you value your sanity. | |
| Apr 1, 2016 at 14:15 | history | edited | Parham Doustdar | CC BY-SA 3.0 |
Completed my example
|
| Apr 1, 2016 at 14:09 | comment | added | Aksakal | excellent question, we need to think more about explaining our concepts to people with disabilities | |
| Apr 1, 2016 at 14:09 | comment | added | Parham Doustdar |
@Antoni Parellada Thanks! I understand what you're saying. What I don't understand is why one would need to use a plot for this. This is something that entirely depends on numbers. We're looking for the formula that has the least amount of errors in estimating Y from X. I understand that perfectly. What I want to understand is how to arrive at the answer without using linear algebra. I'm guessing it is one of the ways to find the answer, but definitely not the only one.
|
|
| Apr 1, 2016 at 14:02 | comment | added | Antoni Parellada | I'll keep on thinking of it, see if I can come up, but right off the bat, think about regression as solving an equation that has no solution. All your data points will be incorrectly predicted by your regressor (the area of the house). You are looking for an equation that makes your errors as tolerable as possible. | |
| Apr 1, 2016 at 13:53 | comment | added | Parham Doustdar | I edited my question to give more context. Does this make things clearer? | |
| Apr 1, 2016 at 13:53 | history | edited | Parham Doustdar | CC BY-SA 3.0 |
Made my question much clearer
|
| Apr 1, 2016 at 13:47 | comment | added | Parham Doustdar | @Antoni Parellada Sorry. My question wasn't clear and I ended up wasting your time. My question is more basic – why? Why a scatter plot? Why can't we just look at the pattern another way? I was assuming that scatter plots are an optional gimmick | |
| Apr 1, 2016 at 13:45 | comment | added | Parham Doustdar |
I'm looking for conceptual, non-mathematical explanations. Although I have studied math in the university, unfortunately most of it has never stuck because of the same reason I'm not understanding linear regression. What I'm not understanding is the reason for using plots and lines for this kind of estimation. It is very strange that one would need to draw a scatter plot to figure out the closest formula that once you give x to it, would spit out y. I'd love to know why things are done the way they are. It's okay if scatter plots are used for presentation purposes, but is that all?
|
|
| Apr 1, 2016 at 13:42 | comment | added | Antoni Parellada | Please let me know if the answer offered is in any way what you were after. If it is off target, I'll be happy to erase it. | |
| Apr 1, 2016 at 13:30 | comment | added | shadowtalker | What's your math background? The Wikipedia page called Simple Linear Regression is mostly text, and has what I think is a reasonably clear description in the first paragraph. How does that article compare to the level of detail you're looking for? | |
| Apr 1, 2016 at 13:27 | comment | added | ttnphns | But do you have spacial imagination that can take over the vision? If yes, I suppose a scatterplot can be imagined some way. I doubt that the essense of regression can be captured by propositional thinking (such as verbal) solely. | |
| Apr 1, 2016 at 12:50 | history | edited | gung - Reinstate Monica |
edited tags
|
|
| Apr 1, 2016 at 12:48 | review | First posts | |||
| Apr 1, 2016 at 12:50 | |||||
| Apr 1, 2016 at 12:48 | history | asked | Parham Doustdar | CC BY-SA 3.0 |