Loading ...
Sorry, an error occurred while loading the content.
 

The mathematics behind Web Analytics?

Expand Messages
  • danford16182
    Hi, I m interested to know what mathematics does a web analyst need to know? For all of you currently working in the Web Analytics field what is the most
    Message 1 of 15 , Dec 11, 2010
      Hi,

      I'm interested to know what mathematics does a web analyst need to know?

      For all of you currently working in the Web Analytics field what is the most common mathematics you come across and have to use on a day to day basis?

      Can anyone provide me a comprehensive list of the maths I should know in order to work in the Web Analytics field?

      Thanks,

      Dan
    • Hussain Ahmed
      HI , Plus , Minus , Average , minimum , Maximum. Having a Mathematical or statistical back group may help in logical thinking however not much practical use of
      Message 2 of 15 , Dec 11, 2010
        HI ,

        Plus , Minus , Average , minimum , Maximum. Having a Mathematical or
        statistical back group may help in logical thinking however not much
        practical use of it within web analytics.

        Regards,

        Hussain


        On Sat, Dec 11, 2010 at 9:02 PM, danford16182 <danford16182@...>wrote:

        >
        >
        > Hi,
        >
        > I'm interested to know what mathematics does a web analyst need to know?
        >
        > For all of you currently working in the Web Analytics field what is the
        > most common mathematics you come across and have to use on a day to day
        > basis?
        >
        > Can anyone provide me a comprehensive list of the maths I should know in
        > order to work in the Web Analytics field?
        >
        > Thanks,
        >
        > Dan
        >
        >
        >


        [Non-text portions of this message have been removed]
      • Daniel Waisberg
        I believe that rather than mathematics, you should focus on statistics. And I have been a few years in the field and the Cartoon Guide to Statistics by Larry
        Message 3 of 15 , Dec 11, 2010
          I believe that rather than mathematics, you should focus on statistics. And
          I have been a few years in the field and the 'Cartoon Guide to Statistics'
          by Larry Gonick, should give you a really good start.
          http://amzn.to/eTuE2E. Worth every dollar.

          On Sat, Dec 11, 2010 at 11:02 PM, danford16182 <danford16182@...>wrote:

          >
          >
          > Hi,
          >
          > I'm interested to know what mathematics does a web analyst need to know?
          >
          > For all of you currently working in the Web Analytics field what is the
          > most common mathematics you come across and have to use on a day to day
          > basis?
          >
          > Can anyone provide me a comprehensive list of the maths I should know in
          > order to work in the Web Analytics field?
          >
          > Thanks,
          >
          > Dan
          >
          >
          >


          [Non-text portions of this message have been removed]
        • lexykassan
          Agreed on the Statistics. That s the majority of the larger projects I get handed. Most typical metrics are simple counts or averages, but when it comes to
          Message 4 of 15 , Dec 12, 2010
            Agreed on the Statistics. That's the majority of the larger projects I get handed. Most typical metrics are simple counts or averages, but when it comes to analyzing what's going on you need slightly more. Tests (Z, T, Chi-Sq), ANOVA, linear regression, multivariate analysis, and logistic regression tend to be the most used, at least where I am.

            --- In webanalytics@yahoogroups.com, "danford16182" <danford16182@...> wrote:
            >
            > Hi,
            >
            > I'm interested to know what mathematics does a web analyst need to know?
            >
            > For all of you currently working in the Web Analytics field what is the most common mathematics you come across and have to use on a day to day basis?
            >
            > Can anyone provide me a comprehensive list of the maths I should know in order to work in the Web Analytics field?
            >
            > Thanks,
            >
            > Dan
            >
          • Yuhui
            Hi Dan, I agree with Hussain and Daniel. At a basic level, you should be comfortable with simple math. Especially percentages, because that s what you ll use
            Message 5 of 15 , Dec 12, 2010
              Hi Dan,

              I agree with Hussain and Daniel. At a basic level, you should be comfortable with simple math. Especially percentages, because that's what you'll use most often, e.g. with KPIs, conversion rates, etc.

              Of course, you'll also need to know why certain metrics are calculated the way they are, e.g. bounce rate is a percentage, and you should know its what (single access / entries) and why (out of everyone who entered, how many puked and left immediately (paraphrasing Avinash Kaushik's favourite quote)?).

              Once you're comfortable with the basic math, you can then "graduate" to statistics, assuming you don't have that background. That introduces a whole new world of thinking and insights. E.g. you can start thinking in terms of correlations, significance, modelling, etc.

              Good luck!

              Regards,
              Yu Hui
              Twitter: @yuhui

              --- In webanalytics@yahoogroups.com, "danford16182" <danford16182@...> wrote:
              >
              > Hi,
              >
              > I'm interested to know what mathematics does a web analyst need to know?
              >
              > For all of you currently working in the Web Analytics field what is the most common mathematics you come across and have to use on a day to day basis?
              >
              > Can anyone provide me a comprehensive list of the maths I should know in order to work in the Web Analytics field?
              >
              > Thanks,
              >
              > Dan
              >
            • Jess Spate
              Hi Daniel, I would look at the mathematics surrounding errors and uncertainties- nothing complex, just learn how propagate +/- error bounds through a simple
              Message 6 of 15 , Dec 12, 2010
                Hi Daniel,

                I would look at the mathematics surrounding errors and uncertainties-
                nothing complex, just learn how propagate +/- error bounds through a simple
                system.

                As long as you are keenly numerate, you probably won't find any of the maths
                involved in everyday analytics too tricky.

                Jess

                On Sun, Dec 12, 2010 at 6:46 AM, Daniel Waisberg <dwaisberg@...>wrote:

                >
                >
                > I believe that rather than mathematics, you should focus on statistics. And
                > I have been a few years in the field and the 'Cartoon Guide to Statistics'
                > by Larry Gonick, should give you a really good start.
                > http://amzn.to/eTuE2E. Worth every dollar.
                >
                > On Sat, Dec 11, 2010 at 11:02 PM, danford16182 <danford16182@...<danford16182%40yahoo.com>
                > >wrote:
                >
                >
                > >
                > >
                > > Hi,
                > >
                > > I'm interested to know what mathematics does a web analyst need to know?
                > >
                > > For all of you currently working in the Web Analytics field what is the
                > > most common mathematics you come across and have to use on a day to day
                > > basis?
                > >
                > > Can anyone provide me a comprehensive list of the maths I should know in
                > > order to work in the Web Analytics field?
                > >
                > > Thanks,
                > >
                > > Dan
                > >
                > >
                > >
                >
                > [Non-text portions of this message have been removed]
                >
                >
                >


                [Non-text portions of this message have been removed]
              • danford16182
                Thanks Lexykassan This is what I was looking for! I ll get learning!
                Message 7 of 15 , Dec 12, 2010
                  Thanks Lexykassan

                  This is what I was looking for! I'll get learning!


                  --- In webanalytics@yahoogroups.com, "lexykassan" <lexykassan@...> wrote:
                  >
                  > Agreed on the Statistics. That's the majority of the larger projects I get handed. Most typical metrics are simple counts or averages, but when it comes to analyzing what's going on you need slightly more. Tests (Z, T, Chi-Sq), ANOVA, linear regression, multivariate analysis, and logistic regression tend to be the most used, at least where I am.
                  >
                  > --- In webanalytics@yahoogroups.com, "danford16182" <danford16182@> wrote:
                  > >
                  > > Hi,
                  > >
                  > > I'm interested to know what mathematics does a web analyst need to know?
                  > >
                  > > For all of you currently working in the Web Analytics field what is the most common mathematics you come across and have to use on a day to day basis?
                  > >
                  > > Can anyone provide me a comprehensive list of the maths I should know in order to work in the Web Analytics field?
                  > >
                  > > Thanks,
                  > >
                  > > Dan
                  > >
                  >
                • Stephane Hamel
                  Hi Dan, here s my 2 cents, I will cross-post to the UBC WA-1 course where you are enrolled. - basic maths, of course - and I would also add rule of three -
                  Message 8 of 15 , Dec 12, 2010
                    Hi Dan,
                    here's my 2 cents, I will cross-post to the UBC WA-1 course where you are enrolled.

                    - basic maths, of course - and I would also add "rule of three"
                    - standard deviation, concept of control limits, outliers (so you understand the significance and validity of the data you analyze)
                    - regression analysis (so you understand correlations)
                    - then, if you do A/B or mutlivariate, some info about those concepts certainly helps

                    Beyond the numbers and calculations, maths & stats are an excellent way to develop your critical thinking and analytical skills.

                    Stéphane Hamel
                    http://immeria.net
                    Tutor, UBC Award of Achievement in Web Analytics

                    --- In webanalytics@yahoogroups.com, "danford16182" <danford16182@...> wrote:
                    >
                    > Hi,
                    >
                    > I'm interested to know what mathematics does a web analyst need to know?
                    >
                    > For all of you currently working in the Web Analytics field what is the most common mathematics you come across and have to use on a day to day basis?
                    >
                    > Can anyone provide me a comprehensive list of the maths I should know in order to work in the Web Analytics field?
                    >
                    > Thanks,
                    >
                    > Dan
                    >
                  • Michael
                    Great Question: Statistics is fundamentally important to web analytics; perhaps more precisely I would describe the math as either applied statistics or
                    Message 9 of 15 , Dec 12, 2010
                      Great Question:

                      Statistics is fundamentally important to web analytics; perhaps more precisely I would describe the math as either applied statistics or econometrics.

                      In terms of econometrics, UBC courses ECON 425-426 are very similar to what I have taken and will cover the material Stéphane mentioned. I apply the lessons learned from those courses on a daily basis.

                      If you would like to pursue this outside of UBC there are excellent free resources available to you. In particular, I can recommend MIT OpenCourseWare as one of the best. Read more on their site : http://ocw.mit.edu/index.htm

                      Using that knowledge takes another set of skills, a set related to programming.

                      Leveraging outside data such as census data is part of econometric research. For this, a scripting language is absolutely essential as data are often messy, need to be pre-processed, and there isn't always going to be someone to do it for you.

                      I like Python: http://python.org/ over Perl http://www.perl.org/

                      After you have the data you need to do some calculations, so you would need to know how to use a statistical programming language. Each school seems to have their, at times transitory, preferred language. My school used Stata which was a gigantic waste of time.

                      R is free, easy to learn, and widely used in web analytics. Download here: http://www.r-project.org/

                      Direct download link of the best R tutorial ever is here: http://cran.r-project.org/doc/contrib/usingR.pdf
                      Note: According to Dr. Maindonald, ANU not allowing R to source the data files directly but you can open the URL and save the page.

                      For the more rigorous, two texts are particularly noteworthy:
                      Modern Applied Statistics with S (Statistics and Computing) - Brand new 4th Edition, R is based on S
                      A Guide to Econometrics - Sixth Edition, although you can find prior editions quite reasonably priced
                      They reside to the right of my mouse.

                      At the recent ACM Data Mining Camp, whose attendees largely consisted of individuals performing research about online behavior, R users made up approximately 99.5% of the overall population. I don't believe that eMetrics has the same rate of R usage . . . but there is still time.

                      A piece of writing about statistics that I am a big fan of is 'The Median Isn't the Message' by Dr. Stephen Jay Gould. I have crated an annotated version to widen the potential audience on my site: http://michaeldhealy.com/sjg/annotated-median-isnt-the-message/

                      Good luck!

                      Michael D. Healy
                      mdh@...
                      http://michaeldhealy.com/
                      http://twitter.com/michaeldhealy
                      --- In webanalytics@yahoogroups.com, "Stephane Hamel" <shamel67@...> wrote:
                      >
                      > Hi Dan,
                      > here's my 2 cents, I will cross-post to the UBC WA-1 course where you are enrolled.
                      >
                      > - basic maths, of course - and I would also add "rule of three"
                      > - standard deviation, concept of control limits, outliers (so you understand the significance and validity of the data you analyze)
                      > - regression analysis (so you understand correlations)
                      > - then, if you do A/B or mutlivariate, some info about those concepts certainly helps
                      >
                      > Beyond the numbers and calculations, maths & stats are an excellent way to develop your critical thinking and analytical skills.
                      >
                      > Stéphane Hamel
                      > http://immeria.net
                      > Tutor, UBC Award of Achievement in Web Analytics
                      >
                      > --- In webanalytics@yahoogroups.com, "danford16182" <danford16182@> wrote:
                      > >
                      > > Hi,
                      > >
                      > > I'm interested to know what mathematics does a web analyst need to know?
                      > >
                      > > For all of you currently working in the Web Analytics field what is the most common mathematics you come across and have to use on a day to day basis?
                      > >
                      > > Can anyone provide me a comprehensive list of the maths I should know in order to work in the Web Analytics field?
                      > >
                      > > Thanks,
                      > >
                      > > Dan
                      > >
                      >
                    • Dave
                      Hi Dan, That s a good question, although I think you may be either surprised or disappointed by the answer. In my experience, the truth is that a spreadsheet
                      Message 10 of 15 , Dec 13, 2010
                        Hi Dan,

                        That's a good question, although I think you may be either surprised or disappointed by the answer. In my experience, the truth is that a spreadsheet package will do most of the heavy lifting for you (although I still frequently refer to a calculator for speed and ease of use). There isn't a great deal of complex maths in basic web analytics, although areas such as A/B testing and multi-variate testing, along with understanding of statistics and statistical significance will benefit from some maths skills.

                        To answer your question directly - conversion rates (x divided by y), percentages (x divided by y as a percentage) and percentage changes (did it go up or down by a% or b%) are key, and used daily. I have some understanding of statistics - my maths background is more chemistry/physics, so I understand mean/mode/median thorougly - but less so on distributions, but I have never found this to be a problem (there are plenty of downloadable spreadsheets out there that will do the maths for me.

                        What you're more likely to benefit from is a combination of an analytical approach to a problem (what's going "wrong" and why) and a creative approach to solving it. In my view, it's not knowing the maths that's important, it's how to wield the maths as a tool to understand the piles of numbers you're going to find yourself generating!

                        Just my thoughts,

                        David

                        --- In webanalytics@yahoogroups.com, "danford16182" <danford16182@...> wrote:
                        >
                        > Hi,
                        >
                        > I'm interested to know what mathematics does a web analyst need to know?
                        >
                        > For all of you currently working in the Web Analytics field what is the most common mathematics you come across and have to use on a day to day basis?
                        >
                        > Can anyone provide me a comprehensive list of the maths I should know in order to work in the Web Analytics field?
                        >
                        > Thanks,
                        >
                        > Dan
                        >
                      • Rod Jacka
                        Further to this post I personally feel that it is very important to have a deeper understanding of statistical theory before applying the tests or techniques
                        Message 11 of 15 , Dec 13, 2010
                          Further to this post I personally feel that it is very important to have a deeper understanding of statistical theory before applying the tests or techniques mentioned (t, chi, z, regression, etc). These are based on assumptions around the shape of the distribution of data which is assumed in most cases to be normally distributed (i.e. the bell curve). In my experience web analytics data is rarely normally distributed. This can lead to inaccurate conclusions and false causality.

                          I am yet to see any of the main web analytics packages provide anything other than simple averages with no descriptions of the make up of the average, e.g. standard error, shape of the distribution (skewness, kurtosis, etc), nor the ability to identify outliers. Hence unless you can get these via exports from a data warehouse and then using tools like SPSS, etc then a working knowledge of basic stats plus good numeracy is good enough. Anyone disagree?

                          My preference for skills is:
                          - good numeracy,
                          - being comfortable that they understand their numbers and importantly the assumptions behind them,
                          - a logical and skeptical mind,
                          - a methodical approach,
                          - the ability to search for multiple explanations or possible solutions to a problem or scenario and to investigate each of these in turn, and
                          - the willingness to prove themselves wrong.

                          Best

                          Rod


                          --- In webanalytics@yahoogroups.com, "lexykassan" <lexykassan@...> wrote:
                          >
                          > Agreed on the Statistics. That's the majority of the larger projects I get handed. Most typical metrics are simple counts or averages, but when it comes to analyzing what's going on you need slightly more. Tests (Z, T, Chi-Sq), ANOVA, linear regression, multivariate analysis, and logistic regression tend to be the most used, at least where I am.
                          >
                          > --- In webanalytics@yahoogroups.com, "danford16182" <danford16182@> wrote:
                          > >
                          > > Hi,
                          > >
                          > > I'm interested to know what mathematics does a web analyst need to know?
                          > >
                          > > For all of you currently working in the Web Analytics field what is the most common mathematics you come across and have to use on a day to day basis?
                          > >
                          > > Can anyone provide me a comprehensive list of the maths I should know in order to work in the Web Analytics field?
                          > >
                          > > Thanks,
                          > >
                          > > Dan
                          > >
                          >
                        • Michael
                          Dan: Statistics is fundamentally important to web analytics; perhaps more precisely I would describe the math as either applied statistics or econometrics. In
                          Message 12 of 15 , Dec 13, 2010
                            Dan:

                            Statistics is fundamentally important to web analytics; perhaps more precisely I would describe the math as either applied statistics or econometrics.

                            In terms of econometrics, UBC courses ECON 425-426 are very similar to what I have taken and will cover the material Stéphane mentioned. I apply the lessons learned from those courses on a daily basis.

                            If you would like to pursue this outside of UBC there are excellent free resources available to you. In particular, I can recommend MIT OpenCourseWare as one of the best. Read more on their site : http://ocw.mit.edu/index.htm

                            Using that knowledge takes another set of skills, a set related to programming.

                            Leveraging outside data such as census data is part of econometric research. For this, a scripting language is absolutely essential as data are often messy, need to be pre-processed, and there isn't always going to be someone to do it for you.

                            I like Python: http://python.org/ over Perl http://www.perl.org/

                            After you have the data you need to do some calculations, so you would need to know how to use a statistical programming language. Each school seems to have their, at times transitory, preferred language. My school used Stata which was a gigantic waste of time.

                            R is free, easy to learn, and widely used in web analytics. Download here: http://www.r-project.org/

                            Direct download link of the best R tutorial ever is here: http://cran.r-project.org/doc/contrib/usingR.pdf
                            Note: According to Dr. Maindonald, ANU not allowing R to source the data files directly but you can open the URL and save the page.

                            For the more rigorous, two texts are particularly noteworthy:
                            Modern Applied Statistics with S (Statistics and Computing) - Brand new 4th Edition, R is based on S
                            A Guide to Econometrics - Sixth Edition, although you can find prior editions quite reasonably priced
                            They reside to the right of my mouse.

                            At the data mining event, whose attendees largely consisted of individuals performing research about online behavior, R users made up approximately 99.5% of the overall population. I don't believe that eMetrics has the same rate of R usage . . . but there is still time.

                            A piece of writing about statistics that I am a big fan of is 'The Median Isn't the Message' by Dr. Stephen Jay Gould. I have crated an annotated version to widen the potential audience on my site: http://michaeldhealy.com/sjg/annotated-median-isnt-the-message/

                            Good luck!

                            Michael D. Healy
                            mdh@...
                            http://michaeldhealy.com/
                            http://twitter.com/michaeldhealy


                            --- In webanalytics@yahoogroups.com, "danford16182" <danford16182@...> wrote:
                            >
                            > Hi,
                            >
                            > I'm interested to know what mathematics does a web analyst need to know?
                            >
                            > For all of you currently working in the Web Analytics field what is the most common mathematics you come across and have to use on a day to day basis?
                            >
                            > Can anyone provide me a comprehensive list of the maths I should know in order to work in the Web Analytics field?
                            >
                            > Thanks,
                            >
                            > Dan
                            >
                          • peter_oneill17
                            Hi Dan, I am finding the split in responses here really interesting, between those people who believe statistics is vital to web analytics and those who rarely
                            Message 13 of 15 , Dec 14, 2010
                              Hi Dan,

                              I am finding the split in responses here really interesting, between those people who believe statistics is vital to web analytics and those who rarely use it. I personally don't use statistics in the vast majority of what I do, instead it is trending and segmenting data, looking for patterns and what has changed. As such, I find numerical reasoning skills to be the most important, something great to pick up is the ability to spot numbers that "look wrong".

                              I do think it is important to have a decent understanding of the principles behind distribution and statistical significance. I did a fair bit of stats at uni and have forgotten most of it but simply knowing what the bell curve represents has been helpful at times (I have now ordered the cartoon guide to statistics for a refresher - thanks Daniel). What I find more dangerous are people who apply standard deviations when it is not a normal distribution (I agree with Rod that this is rarely the case in web analytics) or who claim two metrics are correlated when if they really understood the correlation test they performed, they would know this is incorrect. Excel or other tools can do these calculations for you but it is understanding the principles that is most important.

                              I would not actually worry about what mathematics/statistics you should know, or which programming language, etc. What you need to be able to do is to transform data into insights and to turn these insights into recommendations for improving a business. Different web analysts use different skill sets to achieve these aims, if you hit questions you can't answer, then learn the skill set that will enable you to answer that question.

                              Peter




                              --- In webanalytics@yahoogroups.com, "danford16182" <danford16182@...> wrote:
                              >
                              > Hi,
                              >
                              > I'm interested to know what mathematics does a web analyst need to know?
                              >
                              > For all of you currently working in the Web Analytics field what is the most common mathematics you come across and have to use on a day to day basis?
                              >
                              > Can anyone provide me a comprehensive list of the maths I should know in order to work in the Web Analytics field?
                              >
                              > Thanks,
                              >
                              > Dan
                              >
                            • Jay Tkachuk
                              Hi Dan, A good grasp of basic mathematical and statistical concepts would be helpful for the in-depth analysis and optimization efforts like MVT - statistical
                              Message 14 of 15 , Dec 14, 2010
                                Hi Dan,

                                A good grasp of basic mathematical and statistical concepts would be helpful
                                for the in-depth analysis and optimization efforts like MVT - statistical
                                significance, confidence intervals, correlation, regression analysis, etc.
                                Any Stats 101 textbook would be a good start.

                                Jay


                                [Non-text portions of this message have been removed]
                              • captain_blackbeak
                                The only time I apply statistical analysis is when I have to in order to pinpoint trends, usually in smaller datasets (IE less than a quarter of data) where I
                                Message 15 of 15 , Dec 26, 2010
                                  The only time I apply statistical analysis is when I have to in order to pinpoint trends, usually in smaller datasets (IE less than a quarter of data) where I want to see if anything unusual is going on or large datasets I know should follow a normal bell curve distribution, or we know the growth factor. Generally trends are what we're looking for and most of the time the tools do a good job of pinpointing them for you.

                                  Google already applies statistical analysis to its data in business intelligence. It's basically a great time saver as all its doing is applying statistical deviations to the data over time and telling you what effects the data (when they go beyond the norm).

                                  I wrote an article about the way you can do this in for instance excel if you're interested.
                                  http://www.blackbeak.com/2008/04/16/using-standard-deviations-to-determine-web-analytics-benchmarks/

                                  --- In webanalytics@yahoogroups.com, "peter_oneill17" <updates@...> wrote:
                                  >
                                  > Hi Dan,
                                  >
                                  > I am finding the split in responses here really interesting, between those people who believe statistics is vital to web analytics and those who rarely use it. I personally don't use statistics in the vast majority of what I do, instead it is trending and segmenting data, looking for patterns and what has changed. As such, I find numerical reasoning skills to be the most important, something great to pick up is the ability to spot numbers that "look wrong".
                                  >
                                  > I do think it is important to have a decent understanding of the principles behind distribution and statistical significance. I did a fair bit of stats at uni and have forgotten most of it but simply knowing what the bell curve represents has been helpful at times (I have now ordered the cartoon guide to statistics for a refresher - thanks Daniel). What I find more dangerous are people who apply standard deviations when it is not a normal distribution (I agree with Rod that this is rarely the case in web analytics) or who claim two metrics are correlated when if they really understood the correlation test they performed, they would know this is incorrect. Excel or other tools can do these calculations for you but it is understanding the principles that is most important.
                                  >
                                  > I would not actually worry about what mathematics/statistics you should know, or which programming language, etc. What you need to be able to do is to transform data into insights and to turn these insights into recommendations for improving a business. Different web analysts use different skill sets to achieve these aims, if you hit questions you can't answer, then learn the skill set that will enable you to answer that question.
                                  >
                                  > Peter
                                  >
                                  >
                                  >
                                  >
                                  > --- In webanalytics@yahoogroups.com, "danford16182" <danford16182@> wrote:
                                  > >
                                  > > Hi,
                                  > >
                                  > > I'm interested to know what mathematics does a web analyst need to know?
                                  > >
                                  > > For all of you currently working in the Web Analytics field what is the most common mathematics you come across and have to use on a day to day basis?
                                  > >
                                  > > Can anyone provide me a comprehensive list of the maths I should know in order to work in the Web Analytics field?
                                  > >
                                  > > Thanks,
                                  > >
                                  > > Dan
                                  > >
                                  >
                                Your message has been successfully submitted and would be delivered to recipients shortly.