Which is the best software tool for plotting growth charts?
I want to apply Least Mean Square (LMS) method to a dataset containing hormonal levels related to fertility. My aim is to draw a growth chart similar to one used in Pediatrics. Now the only tool I've found is LMSCHartMaker that is really not useful. Neither other tools such as STATA, SPSS or SAS have this method implemented by default. Can someone please help me? Thanks a lot.
Do you have a model? If you do not (that means the purpose is not to estimate model parameters) you may probably just want to visualize the dynamics by a smooth curve through the data. For this purpose I would recommend to use splines or loess ("locally estimated scatterplot smoothers") curves. In case you will go for a try using R, here an example:
x = runif(50) # simulates the predictor: generates 50 uniformly distributed values between 0 and 1
y = sin(x*4)+rnorm(50,s=0.2) # simulates the response: 50 noisy measurements
plot(y~x) # plots the response values against the predictor values
est.fn = loess(y~x) # creates a loess model that can be used for predictons
x.pred = seq(min(x), max(x), length=100) # 100 values of the predictor at which we like to make a prediction based on the loess-model
lines(y.pred~x.pred) # draw the smooth curve by conneting all predicted points
As there is always more than one way to skin a cat: for instance, using the xyplot function from the library "lattice" does the prediction job in the background (x and y are taken from above):
Theres a really neat package in R called GAMLSS which I used to make birthweight charts but I suspect might be useful here. If I remember correctly it uses the LMS method. The authors of the command are also really helpful to contact with queries.
I believe that Gamlss (R package) is the best tool for this. It includes many methods such as LMS, LMSP, LMST. And, has many distributions to fit. LMSP can also the kurtosis of the distribution, thus fits better than LMS.
R software is the most detailed in some circumstances compared with LMS. But I consider that its wise to use just quartiles if you want to make comparisons. If the range of your figures are wide aritmetic mean can be used
I will use the terms “reference centiles ” or “centile charts, ” as the setting that I consider here is more general than that of “growth charts.” Longitudinal reference centiles over some measure of time (typically age) are almost always implemented repeatedly on the same individual. In this kind of setting the notion of conditional or adaptive ce...