Estimating PDFs with sparse data https://www.lotterypost.com/thread/345335 Lottery Post Forum Topic: Estimating PDFs with sparse data en-us Lottery Post RSS Generator Reply #1 https://www.lotterypost.com/thread/345335/7258003 https://www.lotterypost.com/thread/345335/7258003 Wed, 10 May 2023 21:13:49 GMT eddessaknight Good Post Wavepack

]]> eddessaknight Estimating PDFs with sparse data https://www.lotterypost.com/thread/345335 https://www.lotterypost.com/thread/345335 Wed, 10 May 2023 02:24:06 GMT Wavepack Estimate a multinomial pdf, p t = [p 1 , ... , p m ], of a random variable with m states using N samples of the random variable, given that the prior pdf estimate, p t-1 is a uniform pdf. Denote the sample count as n = [n 1 , ... , n m ].

Why? Because taking a pure frequentest estimation of p n / N, which doesn't take into account prior pdf information, and results in an estimate with too many zero entries, because of sparse data. That's the case when dealing with 2nd order statistics and li... [ More ]

]]> Wavepack