Dynamic model-based clustering for spatio-temporal data

Paci, Lucia; Finazzi, Francesco

doi:10.1007/s11222-017-9735-9

Cited by 13 publications

(15 citation statements)

References 35 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…() employed a two‐state mixture of Gaussian distributions centered at “home” and “work” locations to understand human motion from cell phone data and social networks. More generally, mixture models can be used to represent complex dynamic spatial distributions (Paci and Finazzi, ) and detect relevant places for individuals. However, model estimation is computation demanding and requires estimating the number of clusters for each smartphone user, increasing the computational burden.…”

Section: Methodsmentioning

confidence: 99%

Quantifying Personal Exposure to Air Pollution from Smartphone-Based Location Data

Finazzi

Paci

2019

Biometrics

Self Cite

View full text Add to dashboard Cite

Personal exposure assessment is a challenging task that requires both measurements of the state of the environment as well as the individual's movements. In this paper, we show how location data collected by smartphone applications can be exploited to quantify the personal exposure of a large group of people to air pollution. A Bayesian approach that blends air quality monitoring data with individual location data is proposed to assess the individual exposure over time, under uncertainty of both the pollutant level and the individual location. A comparison with personal exposure obtained assuming fixed locations for the individuals is also provided. Location data collected by the Earthquake Network research project are employed to quantify the dynamic personal exposure to fine particulate matter of around 2500 people living in Santiago (Chile) over a 4-month period. For around 30% of individuals, the personal exposure based on people movements emerges significantly different over the static exposure. On the basis of this result and thanks to a simulation study, we claim that even when the individual location is known with nonnegligible error, this helps to better assess personal exposure to air pollution. The approach is flexible and can be adopted to quantify the personal exposure based on any location-aware smartphone application. K E Y W O R D SDynamic models, Markov chain Monte Carlo, particulate matter, space-time modeling

show abstract

Section: Methodsmentioning

confidence: 99%

Quantifying Personal Exposure to Air Pollution from Smartphone-Based Location Data

Finazzi

Paci

2019

Biometrics

Self Cite

View full text Add to dashboard Cite

show abstract

“…turning-angle) since in the first the wolf tends to move in a straight line while in the second she moves with an anticlockwise pattern. We also show that our model outperforms the HMM and the proposal of [49]. The paper is organized as follows: Section 2 presents the motivating example and a full description of the dataset that will be analyzed, Section 3 describes the proposed approach, Section 4 presents the results of the simulation study while in Section 5 can be seen the results of the model estimated on the real data.…”

Section: Endmentioning

confidence: 99%

“…For instance, the proposal of [40] is obtained by assuming η t ≡ 0 K−1 . The model of [49] is obtained by letting η t be a spatio-temporal process with autoregressive temporal increments, D = (K − 1) and a diagonal matrix A. Moreover, we can reduce to the proposals of [60], [38] and [54] assuming D = 1.…”

Section: The Logistic Normal Approachmentioning

confidence: 99%

“…. , 6} and the same NNGP approximation, with m ∈ {1, 10, 20}, is used for the proposal of [49]. As prior distributions we assume Dirichlet with vectors of parameters equal to (1, .…”

Section: Real Data Examplesmentioning

confidence: 99%

“…It generalizes most of the previous proposals and allows us to perform inference in such a way that it does not depend on the reference element nor on the ordering. Both points have often been overlooked in the literature, see for example [38], [60], [49], and [11]. In other cases, oversimplified hypothesis have been imposed, resulting in a reduced flexibility of the models, e.g.…”

Section: Introductionmentioning

confidence: 99%

See 2 more Smart Citations

New formulation of the logistic-Gaussian process to analyze trajectory tracking data

Mastrantonio¹,

Grazian²,

Mancinelli³

et al. 2019

Ann. Appl. Stat.

View full text Add to dashboard Cite

Improved communication systems, shrinking battery sizes and the price drop of tracking devices have led to an increasing availability of trajectory tracking data. These data are often analyzed to understand animals behavior using mixture-type model.In this work we propose a new model based on the Logistic-Normal process. Due to a new formalization and the way we specify the coregionalization matrix of the associated multivariate Gaussian process, we show that our model, differently from other proposals, is invariant with respect to the choice of the reference element and of the ordering of the components of the probability vectors. We estimate the model under a Bayesian framework, using an approximation of the Gaussian process needed to avoid impractical computational time.We perform a simulation study with the aim of showing the ability of the model to retrieve the parameters used to simulate the data. The model is then applied to the real data where a wolf is observed before and after the procreation. Results are easy interpretable showing differences in the two phases.We record a time series of spatial locations of the female wolf called F24 (2-3 years old), that was live-trapped in May 2009 in the Abruzzo, Lazio and Molise National Park (central Apennines, Italy) and equipped with a Vectronic Pro Light-1 collar (Vectronic Aerospace GmbH, Berlin, Germany). Details on wolf capture and handling are provided in [36]. During winter months (January -April), fix attempts were scheduled every 30 minutes for 10 days, and every 3 hours for 20 days for the rest of the year. The lower acquisition interval during winter months was programmed with the aim to estimate wolf kill rate through field investigations of GPS clusters [58].When first captured, F24 was a member of the Villa pack, where it remained for 7.9 months before dispersing and establishing a new pack (i.e., Bisegna pack) in January 2010. The wolf reproduced in May 2010 and, using information derived from its GPS locations, we were able to determine the actual position of the den. Between May 28th and June 4th, F24 restricted its movements in the proximity of the den. We therefore assumed that F24 entered the den on May 28th and reproduced in the following days. Until collars failure on June 16th, F24 systematically revisited the den to feed and attend the cubs. We collected locations from F24's collar with a mean acquisition rate of 91.5%. We decided to The modelIn a mixture-type model the clustering is generally encoded using a discrete (latent) random variable z = {z t } t∈T , where z t ∈ {1, 2, . . . , K} ≡ K is a membership variable such that z t = k indicates the behavior at time t.

show abstract

Spatiotemporal clustering using Gaussian processes embedded in a mixture model

2021

View full text Add to dashboard Cite

The categorization of multidimensional data into clusters is a common task in statistics. Many applications of clustering, including the majority of tasks in ecology, use data that is inherently spatial and is often also temporal. However, spatiotemporal dependence is typically ignored when clustering multivariate data. We present a finite mixture model for spatial and spatiotemporal clustering that incorporates spatial and spatiotemporal autocorrelation by including appropriate Gaussian processes (GP) into a model for the mixing proportions. We also allow for flexible and semiparametric dependence on environmental covariates, once again using GPs. We propose to use Bayesian inference through three tiers of approximate methods: a Laplace approximation that allows efficient analysis of large datasets, and both partial and full Markov chain Monte Carlo (MCMC) approaches that improve accuracy at the cost of increased computational time. Comparison of the methods shows that the Laplace approximation is a useful alternative to the MCMC methods. A decadal analysis of 253 species of teleost fish from 854 samples collected along the biodiverse northwestern continental shelf of Australia between 1986 and 1997 shows the added clarity provided by accounting for spatial autocorrelation. For these data, the temporal dependence is comparatively small, which is an important finding given the changing human pressures over this time.

show abstract

Dynamic model-based clustering for spatio-temporal data

Cited by 13 publications

References 35 publications

Quantifying Personal Exposure to Air Pollution from Smartphone-Based Location Data

Quantifying Personal Exposure to Air Pollution from Smartphone-Based Location Data

New formulation of the logistic-Gaussian process to analyze trajectory tracking data

Spatiotemporal clustering using Gaussian processes embedded in a mixture model

Contact Info

Product

Resources

About