I'm doing statistics on the time during a car trip (only one gas tank,
no refill) and I supposed a lognormal distribution because the
probability of getting a negative time is 0. With the data I've got,
the distribution looks like lognormal, but a little bit on the right.
I tought it comes from the fact that one gas tank lasts around 10
hours. Is there any distribution that looks like lognormal (or
gaussian) I could use to get P(T > 0) = P(T < 10) = 0?
Thank you.
I assume you mean P(T < 0) = P(T > 10) = 0. Have you tried fitting a
scaled Beta: f[t|a,b,c] = (t/c)^(a-1) (1 - t/c)^(b-1) / (c Beta[a,b]),
which has P(T < 0) = P(T > c) = 0.