[R] Goodness of Fit for Word Frequency Counts
Dear Mailing list! sorry to bother you - but maybe you can help me out. I have been searching and searching for appropriate tests. I have a huge dataset of loan requests and I have data at portfolio level, with average portfolio size 200 loans. I want to test whether portfolios are randomly drawn. The problem is that I have rather qualitative data, namely I want to characterize whether loans are randomly selected using word counts. For each loan, I have a "sector", "activity" and "use description". The "use description" contains about 15 words, the "activity" description is usually only one or two words. What I did as of now was to find the word-counts in the overall portfolio, which is 110,000 loans. From this, I can compute, based on knowing the size of a team portfolio the expected frequency of certain keywords appearing. The "sector" variable is categorical and can take only 17 values, whereas in the overall distribution I found 180 different words as activity description. I now wanted to do a type of "goodness of fit" test to see whether the portfolios are randomly selected or not. I would expect that certain portfolios are indeed randomly selected, whereas others arent. I did a chi^2, Pearson Tukey and G-test of Goodness of Fit. The problem is that these tests are usually constructed for categorical data - but if I use the "activity" word-count it need not be categorical. So I am wondering, whether this is still appropriate? I may have a portfolio of 200 loans in which certain words never appear. In this case, I am not sure which degrees of freedom to look at. Should I use as prescribed 179 degrees of freedom as I have 180 "categories" - but these "arent" real categories... An example may look as follows - the word is on the left, the expected and observed word counts are given: +---+ |word observed expected | |---| 1. | food 54 57.511776 | 2. | retail 4649.04432 | 3. |agriculture 39 36.557732 | 4. | services 23 15.867387 | 5. | clothing 13 14.126975 | |---| 6. | transportation 10 6.5851929 | 7. |housing 3 4.65019 | 8. | construction 2 4.3173841 | 9. | arts 5 4.2500955 | 10. | manufacturing 1 3.0170768 | |---| 11. | health 21.751323 | 12. |use 0 .55215646 | 13. | personal 0 .25241221 | 14. | education 2 .68743521 | 15. | wholesale 0 .11241227 | |---| 16. | entertainment 0 .32743521 | 17. | green 0 .42743782 | +---+ I can have R calculate the ChiĀ² statistic from this, but should I use now 17 degrees of freedom? The problem is this is not categorical data! In this case, do I have to make comparisons on a word-by-word basis? Like a "Bernoulli"? I was looking for other goodness of fit tests for this kind of data for days now, but I cant really find any others! I really appreciate your thoughts, best Thiemo --- http://freigeist.devmag.net __ R-help@r-project.org mailing list https://stat.ethz.ch/mailman/listinfo/r-help PLEASE do read the posting guide http://www.R-project.org/posting-guide.html and provide commented, minimal, self-contained, reproducible code.
[R] R book for economists
Dear Group, I am an economics student starting with PhD work in London. As preparation I would like to get to know R a little bit better. For Stata there are tons of books, however, can you recommend a book for R? I have some substantiated econometrics knowledge, so it should be more a how-to book. Best regards Thiemo --- Thiemo Fetzer, Economist http://freigeist.devmag.net http://www.devmag.net __ R-help@r-project.org mailing list https://stat.ethz.ch/mailman/listinfo/r-help PLEASE do read the posting guide http://www.R-project.org/posting-guide.html and provide commented, minimal, self-contained, reproducible code.
Re: [R] Regression inclusion of variable, effect on coefficients
Hello! I was thinking again about the possible interaction between x1 and x4. Theoretically it makes sense, that the influence of x4 on y is the stronger, the less informative is x1. It can be argued that the higher x1, the less informative it is x1. How could I incorporate this relationship in the model? Thanks a lot for your help in advance, Thiemo -Original Message- From: Uwe Ligges [mailto:[EMAIL PROTECTED] Sent: Montag, 21. April 2008 18:54 To: Thiemo Fetzer Cc: r-help@r-project.org Subject: Re: [R] Regression inclusion of variable, effect on coefficients This is not a dump question. This is a serious problem and it depends on what you know or assume about the relastionship between x1 and x4. If you assume linear interaction, you might want to introduce some interaction term to the model for example. Uwe Ligges Thiemo Fetzer wrote: > Hello dear R users! > > I know this question is not strictly R-help, yet, maybe some of the guru's > in statistics can help me out. > > > > I have a sample of data all from the same "population". Say my regression > equation is now this: > > > > m1 <- lm(y ~ x1 + x2 + x3) > > > > I also regress on > > > > m2 <- lm(y ~ x1 + x2 + x3 + x4) > > > > The thing is, that I want to study the effect of "information" x4. > > > > I would hypothesize, that the coefficient estimate for x1 goes down as I > introduce x4, as x4 conveys some of the information conveyed by x1 (but not > only). Of course x1 and x4 are correlated, however multicollinearity does > not appear to be a problem, the variance inflation factors are rather low > (around 1.5 or so). > > > > I want to basically study, how the interplay between x1 and x4 is, when > introducing x4 into the regression equation and whether my hypothesis is > correct; i.e. that given I consider the information x4, not so much of the > variation is explained via x1 anymore. > > > > I observe that introducing x4 into the regression, the coefficient estimate > for x1 goes down; also the associated p-value becomes bigger; i.e. x1 > becomes comparatively less significant. However, x4 is not significant. Yet, > the observation is in line with my theoretical argument. > > > > The question is now simple: how can I work this out? > > > > I know this is likely a dumb question, but I would really appreciate some > links or help. > > > Regards > > Thiemo > > > [[alternative HTML version deleted]] > > __ > R-help@r-project.org mailing list > https://stat.ethz.ch/mailman/listinfo/r-help > PLEASE do read the posting guide http://www.R-project.org/posting-guide.html > and provide commented, minimal, self-contained, reproducible code. __ R-help@r-project.org mailing list https://stat.ethz.ch/mailman/listinfo/r-help PLEASE do read the posting guide http://www.R-project.org/posting-guide.html and provide commented, minimal, self-contained, reproducible code.
Re: [R] Regression inclusion of variable, effect on coefficients
Hello :) I am happy to hear that I am not necessarily asking stupid questions. The thing is, that I have data on x1 and x4 for the whole sample. However, theoretically, it is clear that the informational content of x1 is not as high as of x4. x4 provides more accurate information to the subjects participating in the game, as it has been experimentally and theoretically shown that the x1 is biased. So the experimentators introduced x4 in response to the biased x1. Both prevail however together, so that the subjects have available information on x1 and x4. Theoretically, I argued that the "relative importance" of x1 on y will decrease in light that information x4 is available, as x4 is more accurate. With a simple regression, however, I do not find significant relationships. For x1 it has been empirically and theoretically shown that it has a positive effect on y. The same should hold for x4. There is no necessary theoretical argument as how x1 and x4 interact mathematically, as they both are a measure of the same thing. Yet, x4 is more accurate and contains even more information. It could be any kind of interaction. They are positively correlated, which is also reasonable. Could you suggest me a simple interaction model, with which I could try my luck? Thanks a lot Thiemo -Original Message- From: Uwe Ligges [mailto:[EMAIL PROTECTED] Sent: Montag, 21. April 2008 18:54 To: Thiemo Fetzer Cc: r-help@r-project.org Subject: Re: [R] Regression inclusion of variable, effect on coefficients This is not a dump question. This is a serious problem and it depends on what you know or assume about the relastionship between x1 and x4. If you assume linear interaction, you might want to introduce some interaction term to the model for example. Uwe Ligges Thiemo Fetzer wrote: > Hello dear R users! > > I know this question is not strictly R-help, yet, maybe some of the guru's > in statistics can help me out. > > > > I have a sample of data all from the same "population". Say my regression > equation is now this: > > > > m1 <- lm(y ~ x1 + x2 + x3) > > > > I also regress on > > > > m2 <- lm(y ~ x1 + x2 + x3 + x4) > > > > The thing is, that I want to study the effect of "information" x4. > > > > I would hypothesize, that the coefficient estimate for x1 goes down as I > introduce x4, as x4 conveys some of the information conveyed by x1 (but not > only). Of course x1 and x4 are correlated, however multicollinearity does > not appear to be a problem, the variance inflation factors are rather low > (around 1.5 or so). > > > > I want to basically study, how the interplay between x1 and x4 is, when > introducing x4 into the regression equation and whether my hypothesis is > correct; i.e. that given I consider the information x4, not so much of the > variation is explained via x1 anymore. > > > > I observe that introducing x4 into the regression, the coefficient estimate > for x1 goes down; also the associated p-value becomes bigger; i.e. x1 > becomes comparatively less significant. However, x4 is not significant. Yet, > the observation is in line with my theoretical argument. > > > > The question is now simple: how can I work this out? > > > > I know this is likely a dumb question, but I would really appreciate some > links or help. > > > Regards > > Thiemo > > > [[alternative HTML version deleted]] > > __ > R-help@r-project.org mailing list > https://stat.ethz.ch/mailman/listinfo/r-help > PLEASE do read the posting guide http://www.R-project.org/posting-guide.html > and provide commented, minimal, self-contained, reproducible code. __ R-help@r-project.org mailing list https://stat.ethz.ch/mailman/listinfo/r-help PLEASE do read the posting guide http://www.R-project.org/posting-guide.html and provide commented, minimal, self-contained, reproducible code.
[R] Regression inclusion of variable, effect on coefficients
Hello dear R users! I know this question is not strictly R-help, yet, maybe some of the guru's in statistics can help me out. I have a sample of data all from the same "population". Say my regression equation is now this: m1 <- lm(y ~ x1 + x2 + x3) I also regress on m2 <- lm(y ~ x1 + x2 + x3 + x4) The thing is, that I want to study the effect of "information" x4. I would hypothesize, that the coefficient estimate for x1 goes down as I introduce x4, as x4 conveys some of the information conveyed by x1 (but not only). Of course x1 and x4 are correlated, however multicollinearity does not appear to be a problem, the variance inflation factors are rather low (around 1.5 or so). I want to basically study, how the interplay between x1 and x4 is, when introducing x4 into the regression equation and whether my hypothesis is correct; i.e. that given I consider the information x4, not so much of the variation is explained via x1 anymore. I observe that introducing x4 into the regression, the coefficient estimate for x1 goes down; also the associated p-value becomes bigger; i.e. x1 becomes comparatively less significant. However, x4 is not significant. Yet, the observation is in line with my theoretical argument. The question is now simple: how can I work this out? I know this is likely a dumb question, but I would really appreciate some links or help. Regards Thiemo [[alternative HTML version deleted]] __ R-help@r-project.org mailing list https://stat.ethz.ch/mailman/listinfo/r-help PLEASE do read the posting guide http://www.R-project.org/posting-guide.html and provide commented, minimal, self-contained, reproducible code.
[R] Format regression result summary
Hello to the whole group. I am a newbie to R, but I got my way through and think it is a lot easier to handle than other software packages (far less clicks necessary). However, I have a problem with respect to the summary of regression results. The summary function gives sth like: Residuals: Min 1Q Median 3Q Max -0.46743 -0.09772 0.01810 0.11175 0.42252 Coefficients: Estimate Std. Error t value Pr(>|t|) (Intercept) 3.750367 0.172345 21.761 < 2e-16 *** Var1 -0.002334 0.009342 -0.250 0.802948 Var2 0.012551 0.005927 2.117 0.035444 * Var3 0.015380 0.074537 0.206 0.836730 Var3 0.098602 0.026448 3.728 0.000250 *** ... Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1 Residual standard error: 0.1614 on 202 degrees of freedom Multiple R-squared: 0.1983, Adjusted R-squared: 0.1506 F-statistic: 4.163 on 12 and 202 DF, p-value: 7.759e-06 However, my wish is the output to have a format like: Estimate (Intercept) 3.750367*** (0.172345) Var1 -0.002334 (0.009342) Var2 0.012551* (0.005927) Etc. so that the standard errors are in parantheses below the estimates. Next to the estimates should be the * indicating significance. I thought that should go by accessing the elements in the summary object, yet, I got started and figured that is quite complicated. Is there a quick and dirty way? Basically I want the same print-out as the summary, except that I don't want the t-statistic and not the p-value, only the significance codes. Thanks a lot in advance Thiemo __ R-help@r-project.org mailing list https://stat.ethz.ch/mailman/listinfo/r-help PLEASE do read the posting guide http://www.R-project.org/posting-guide.html and provide commented, minimal, self-contained, reproducible code.