The mixtures are partitioned (note "pi" is a mixing coefficient), so that the red curve should integrate to 1.
p ----- Original Message ----- From: "Colin Rose" <[EMAIL PROTECTED]> To: <[EMAIL PROTECTED]> Sent: Tuesday, May 04, 2004 8:15 AM Subject: Re: [edstat] Mixture of normal distributions > Juha Puranen <[EMAIL PROTECTED]> wrote: > > > A Flash demo for this > > http://noppa5.pc.helsinki.fi/koe/flash/distrib/mixture7.html > > > Neat. But the red curve here does not appear to be a valid pdf -> it > will not integrate to unity (assuming the components are valid). > > > Cheers > > Colin > . > . > ================================================================= > Instructions for joining and leaving this list, remarks about the > problem of INAPPROPRIATE MESSAGES, and archives are available at: > . http://jse.stat.ncsu.edu/ . > ================================================================= > . . ================================================================= Instructions for joining and leaving this list, remarks about the problem of INAPPROPRIATE MESSAGES, and archives are available at: . http://jse.stat.ncsu.edu/ . =================================================================
