I have 5000 sample data recorded from real experiments. From that data how can we know which statistical model does it follow? That means, if we have given sample data, is it possible to develop statistical model from that sample so that we can say the given sample data follows XYZ distribution? Any ideas?
The first step might be to plot the histogram or density plot. Than you can have an idea which theoretical distribution it might follow (normal, lognormal, weibull...). Than you can fit that distribution and test wether your data really fits to that distribution using e.g. the Kolmogorv-Smirnoff Test.
This kind of analysis should be possible in all kind of statistical software, here a tutorial how to do with R: ftp://cran.r-project.org/pub/R/doc/contrib/Ricci-distributions-en.pdf.
Mark,
Could you explain more about the independent and dependent variables you have?
You can always establish a relationship between different types of measurements.
The first step might be to plot the histogram or density plot. Than you can have an idea which theoretical distribution it might follow (normal, lognormal, weibull...). Than you can fit that distribution and test wether your data really fits to that distribution using e.g. the Kolmogorv-Smirnoff Test.
This kind of analysis should be possible in all kind of statistical software, here a tutorial how to do with R: ftp://cran.r-project.org/pub/R/doc/contrib/Ricci-distributions-en.pdf.
- Badges
- Science topic
- Similar topics
- Mathematics
- Statistics
- Statistical Modeling