> IMHO plotting the distribution should be the first step before trying to compute its statistics. If you know the shape, you can understand the values - otherw

> IMHO plotting the distribution should be the first step before trying to compute its statistics. If you know the shape, you can understand the values - otherwise it's guesswork. My manager's dictum is histogram and time series to start, which falls under the same auspices.