Reading a great book twice or even multiple times is a common habit of mine.
R for Marketing Research and Analytics is such a book.
You can read the first 142 pages free here.
The book teaches R in a nice and motivational way. After a general intro it applies R to the simulation of data that is then used to show Marketing Analytics.
Great approach. What I like the most are the many practical tips. E.g. it develops 7 different linear models and does so in a logical way. Then it summarizes all steps nicely at the end of each chapter for quick review.
There is now also the - nearly - same book available for Python: Python for Marketing Research and Analytics.
The R book shows nicely why R is still the better choice for researchers as it provides great sophisticated commands and libraries to get very fast insights.
The plot below required only these two lines of code to check out your m1 model:
par(mfrow=c(2,2))
plot(m1)
This is a overall satisfaction linear model for an amusement park.
sat.df <- read.csv("http://r-marketing.r-forge.r-project.org/data/rintro-chapter7.csv")
Since distance (to the park) is skewed, log has been applied and then lm() does the linear modeling. It produces an object that can be used with plot(), summary(), predict(), and other functions to inspect the model fit and estimates of the coefficients.
m1 is a very simple model as it takes only the rides satisfaction variable.
sat.df$logdist <- log(sat.df$distance)
m1 <- lm(overall ~rides, data = sat.df)
The final model m7 contains one interaction term, wait:has.child, and many more variables plus the data has been normalized (deduct means and divide by standard deviations).
m7 <- lm(overall ~ rides + games + wait + clean + logdist + has.child + wait:has.child, data = sat.std)
Try that with Python. I will get the Python book btw today and maybe I change my opinion then a bit.
