0% found this document useful (0 votes)
3 views1 page

Recidivism Modeling and Estimation Techniques

Uploaded by

lujiahe2001
Copyright
© All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd
0% found this document useful (0 votes)
3 views1 page

Recidivism Modeling and Estimation Techniques

Uploaded by

lujiahe2001
Copyright
© All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd

MIE 360: Systems Modeling and Simulation (Fall 2023)

Tutorial 4: Input Modelling: Parameter Estimation

1. Los Angeles County, home to the world’s largest jail system, would like to have a better
prediction of its ever growing jail population. A major concern for the authorities is to have
an accurate model of recidivism, i.e., the tendency to commit another crime upon release.
As a step, they would like to model the time released recidivists spend on the streets before
being rearrested for their next offense.
Attached is a list of 1,000 released recidivists in 2012 with their rearrest dates. Literature
suggests that the lognormal distribution is a very accurate distribution model to reflect
recidivism duration. A repeat recidivist’s story on the streets typically consists of three
stages: a brief period of desistance, re-criminalization, and evasion. The duration of each
part are almost in proportion to each other.

(a) For the given data, compute the method of moments and maximum likelihood esti-
mators of µ and σ. Hint: (1) You might want to consider the general case with N
(ln(x)−µ)2
data points; (2) The lognormal pdf is xσ
1


e− 2σ 2 ; (3) If X ∼ Lognormal(µ, σ),
2 2
kµ+ k 2σ
E(X k ) = e .

2. You suspect that the completion time of a certain activity in a large construction project
follows a distribution with probability density function (p.d.f.) and cumulative distribution
function (c.d.f.) given respectively by:
( (
a
(1+x)a+1
, x ≥ 0, 1 − (1 + x)−a , x ≥ 0,
f (x) = and F (x) =
0, x < 0, 0, x < 0.

(a) Compute the Method of Moments (MoM) estimator for parameter α, given Pn100= 100
historical observations of the duration of the activity X1 , X2 , . . . , X100 with i=1 Xi =
120.5.
(b) Denote by τ the amount of time such that with 90% probability the duration of the
activity is smaller than τ . Compute τ , assuming that the duration of the activity is
distributed according to F (x).

You might also like