Comments on: Linear regression analysis in Excel

Run regression analysis in Excel to get the answers to the following questions: Which factors matter and which can be ignored? How closely are these factors related to each other? And how certain can you be about the predictions? Continue reading →

Comments page 5. Total comments: 155

  1. Thank you very much. The best explanation I've found. Wonderful!!!

  2. Hi Svetlana. I never place my comments but your tutorial is worth it! All other pages either just show how its done or explain it very "statistically". But you explained it like teaching to a child. Thank you again for this!

  3. Best explanation of regression ever. Very simple, clear and easy to understand. Thank you so much and will keep a tab on your tutorials.

  4. thanks a lot. really helped.

  5. Very well explained! Thanks.

  6. Hi!
    Also for me it was really helpful!
    Unfortunately I cannot produce a graph for a multiple linear regression. Is Excel not able to show it or do you have a tutorial about it, which could help me?
    Thanks!

  7. This write-up has really helped me, but I'm left with one other question:
    If I am using linear regression on a standard curve (say response over concentration) to obtain an equation that I can use to determine the concentration of unknown samples, how do I determine the uncertainty of the concentration value that this equation yields?

    Thank you for this tutorial, it was a lifesaver!

  8. I agree with Andre above. this is the clearest tutorial ever. So easy to follow.
    Further question, how do you deal with blank spaces in the data? If I chart it, I can still get a trend line and equation but if I try to work directly from the data it balks.
    Thank you very much

  9. Wow, first excel tutorial I read that is clear from A to Z...nice!

  10. Hi Svetlana,
    I really like your explanations for linear regression, but I am confused about your explanation on Significance F value. In the example you provided, you explained that If Significance F is less than 0.05 (5%), your model is OK. If it is greater than 0.05, you'd probably better choose another independent variable.
    But When responding to Ali's question whose Significance F value is 6.07596E-31, you said " in your case, Significance F is far less than 5%, so your results are statistically significant." So between (0.05) and (6.07), which one is greater than the other one?
    I have run my regression and my Significance F value is 0.005590647. Is this value (0.005) greater or less than (0.05)?
    Please, I am confused.

    1. Hi!

      6.07 is of course greater than 0.05. However, 6.07596E-31 is not 6.07, it is 0.000000000000000000000000000000607596! 6.07596E-31 is a special format (Scientific notation) used by Excel to display very large and very small numbers in a compact way. When responding to Ali's question I briefly explained about the Scientific format, you can find more info here: Scientific notation format in Excel.

      In your case, Significance F (0.005590647) is also less than 0.05 - the more zeros after the decimal point the smaller the number.

  11. Simple and well explained!

  12. Very helpful for uni thank you very much - so much information which really helped and explained things I could not find the answers to anywhere but here, many thanks :)
    bob the bacteria

  13. Wrong information given here:
    Written 'If you use two or more explanatory variables to predict the independent variable, you deal with multiple linear regression.'
    It is Not the independent variable you predict. It is the dependent variable.

    1. Hi!
      Of course, you predict the dependent variable. Sorry for that typo, fixed.

  14. It was very helpfuk to me and easy to learn method

  15. Very well explained

  16. Awesome!

  17. Indeed very detailed and helpful. Thanks so much

  18. Thanks a lot, so easy to understand

  19. Very helpful, so clear to understand though I've already studied statistic for several weeks. I've already saved it. Thanks very much!

  20. Very helpful. Thanks for that.

  21. Thanks for this.
    Any link for Logistic Regression?

  22. Thank you very much for the information . My assignment became very easy and understandable through the information provided here. It gave me the conceptual clarity.
    Heaps of thanks!

  23. Hi Mam,
    I find it extremely helpful for my M.Tech project where i have to perform regression analysis. But can you please tell me if we can use this LINEST function can be implemented for lognormal equations as ln(y)=a+bln(x)....

  24. Thank you so much @ Svetlana. It's been really helpful

  25. The tutorial was easy to understand and was also helpful. I am no longer a novice as far as regression analysis is concern. Thanks.

  26. Hi Svetlana,
    Thanks for your detailed and well-written article on regression in the Analysis Toolpak and also for mentioning RegressIt as a professional-grade alternative. I'd like your readers to know that RegressIt is free (unlike XLSTAT) and has many features that are designed to help users learn and apply best practices of regression modeling. Variables are selected from a list of names (rather than by entering coordinates of cell ranges), and there are tools for testing model assumptions, comparing models side by side, and sharing results with others in presentation quality format. The graphs and tables that it produces are far superior to what you get with the Analysis Toolpak, both in terms of design and in terms of the set of options that are available. RegressIt also includes very detailed built-in teaching notes that can be embedded in the model worksheets, and it has features that help instructors to grade and verify the originality of work submitted by students. Also, unlike the Analysis Toolpak, it has the capability to forecast from a regression model (including an option for interactive confidence limits on forecast charts), which addresses the question raised by one of your respondents (Syed, post #34). And one more thing: it includes a user-friendly interface with R that allows users to run both linear and logistic regression models in R without writing any code. This feature allows more sophisticated model testing to be carried out and provides a gentle introduction to the R environment for those without programming experience. I encourage you to mention some of these features to your readers, either in this list of comments or a separate post.
    Cheers,
    --Bob

  27. Thanks a lot ma'am

  28. Hi Svetlana,

    Your post is amazing; it must have helped millions including me.
    One simple confusion; now i can i forecast using this regression concept. So if have to predict future, i can use this concept.can you take one example.
    Many thanks in advance.

  29. So helpful. Thanks!

  30. THIS IS EXCELLENT, IT HAS HELP ME DOING MY PROJECT.

  31. This was so useful.. Thanks a ton

  32. I have two variables say time and one predictor variable. However, the predictor variable is classified into regions. Is there a possibility of having that captured in a regression output as a grouping variable in excel?

  33. Very helpful. Detailed and clear explanation.

  34. Thank you so much

    Please also send the link. How to do multiple regression, non-linear regression

    With your best Regards
    Tanveer

  35. Great explanation, much appreciated Excel Functions there!

  36. Thanks a lot.

  37. Mrs. Svetlana,
    Congratulations for great work on this topic.
    You made it easy to understand in short time.
    BRAVO

  38. Really helpful and easy to understand. Thanks.

  39. Thanks a lot! It`s very interesting and useful!

  40. Thanks a lot, it was too useful!

  41. my Significance F value is 6.07596E-31
    what does it mean?

    1. Hi Ali,
      It is a scientific notation that replaces part of the number with E+n, where E (exponent) multiplies the preceding number by 10 to the nth power. That is, 6.07596E-31 equals 6.07596x10^-31 (6.07596 times 10 to the -31st power).

      The Significance F value measures the reliability of the results. If it is less than 0.05 (5%), your model is OK. In your case, Significance F is far less than 5%, so your results are statistically significant.

      1. Hey! Can you also provide this data?

        1. Hey Mona, what data hey ! You send data 15000.

        2. Around 500 enteries if you have! please upload that too

  42. Very clear, helped me a ton. Thank you so much :)

  43. Am definitely getting an A with the explanation of these awesome work. Thank you

  44. that was really really helpful.

  45. Highly informative. I love that. God blesses you.

  46. Hi Svetlana,
    Your article is very nice and its self explanatory for beginners like me.

  47. so helpful! will be using this site more often

  48. do you have any post about what is difference between standardized versus non-standardized coefficient? i don't get it

  49. Why does # of observations equal # data points minus 1? It is not immediately apparent in your example because you include the title rows in your input ranges.
    Besides this one issue the article is great and extremely informative

    1. Hi Braden,
      It's a very good question.
      I included the header row in the input ranges to make it easier to interpret the regression analysis output, based on the column headers. The number of observations equals the number of data points (24 observations, rows 2-25); the header row is not counted because the Labels box was selected.

  50. Very good notes

Post a comment



Thank you for your comment!
When posting a question, please be very clear and concise. This will help us provide a quick and relevant solution to
your query. We cannot guarantee that we will answer every question, but we'll do our best :)