deletions | additions       diff --git a/Approach.tex b/Approach.tex  index 4051a0d..caecf56 100644  --- a/Approach.tex  +++ b/Approach.tex 
...
\[  \widehat{\lambda} = \frac{1}{n}\sum_{i=1}^n k_i.  \] Perhaps it is a good idea to use maximum likelihood estimation to fit the estimates from the 4 (or 3, if we want a hold-out set) to a normal distribution $\lambda\sim\mathcal{N}\left(a,b\right)$ so that we can get a confidence interval for the value of $\lambda$ at each time of day. There are also papers out there which talk about more advanced methods of fitting an inhomogeneous Poisson process. This one has a nice overview: http://www.orsj.or.jp/~archive/pdf/e_mag/Vol.42_4_471.pdf