Is this a suitable scale for a linear regression line? The minimum of the mean weight is 0.1275 and the maximum is 11.6140.
It just seems far too big for how small the minimum is. Minitab wont let me show the exact values for the mean for some reason
>>7719494
why are you using a linear fit on something that appears to be exponential?
>>7719516
We were told a liner regression line was meant to be used. We were given the weights of fish caught on specific dates in the year.
I decided to make a scatter plot of the data of the mean rather than the all of the data because it looks like a mess.
>>7719522
so you were told to use linear regression and I guess you used some sort of built in matlab function to get the regression line....
why are you doubting your result?
>>7719558
Because of how retarded it is, we're then told to find the y intercept and the slope, the Y intercept comes out as 0.02518 (which the *linear* regression line clearly shows is wrong) and as for the slope I can't find a value as it's supposed to give me.
Basically we were given data by Jan 3, March 7 etc.etc. and each date had around 60+ weights for fish, we're supposed to plot that one a graph and get a regression line but you can't use date as an independant variable for a scatter plot, it doesn't work.
So I changed the individual dates to the day# of the year, (so jan 1st is the first day of the year, etc.etc.) and its a complete cluster fuck.
>>7719562
No the linear regression is probably the best it can be. If it looks so wrong to you, it's because using a linear regression is retarded in this case.
Just give the y intercept of the line and leave it at that :)
>>7719611
Is there nothing I can do with my data to make it better or is whatever I'm doing fine?
>>7719614
You can make a linear regression on ln(data) I guess? You might get better results
>>7719618
I don't think that's what he wants us to do he never went over that. it wouldn't be so bad if this wasn't worth 15% of the class but it is, I emailed him 2 weeks ago and haven't had a reply
>>7719494
>fitting a linear function to data that obviously shows exponential behavior
>>7719614
OP I guess the idea is to show you that the applicability or correctness of linear regression is dependent on the behavior of the actual data...
nontheless.. you would be surprised how many times linear regression is used to side step other non linear behaviours
>>7719646
hey I don't make the rules, I just follow them.
>>7719650
That may be true. thanks I guess. I just don't see why they would ask for this kind of data set to be on a scatterplot.
as illustrated in >>7719522 the way that the data has been given is stupid, and as far as I'm aware there is no other type of graph suitable for this where I can fit a linear regression line. for the mean values I used a bar graph because obviously thats the best one, but after being asked to transform the data changing it the date to the day in the year was the best I could do to make it work.
bump?
>>7719655
There is most likely some sort of educational purpose here. A scatterplot makes for easy identification of nonlinear relations. Dont overthink it.
>>7720300
but it doesn't make any sense for the further questions, that's why I'm losing my mind. we need to know the intercept and the slope, but it doesn't display the value of the slope because it seem date is a really shitty x value.
>>7721648
You have a slope right there. You're missing the intercept, probably because you accidentally forced the intercept to be zero. Least squares is just a minimization of a quadratic function and can't ever fail to return results. Bad data will just make the results lose meaningful interpretation.
>>7721699
3.57 is wrong though, theres no way that's the right value
