Doctor-prescription amounts in the Midi-Pyrenees Department in 1999

Description:

These data, analysed in Cressie, Perrin, and Thomas-Agnan (2005) and Cressie and Wikle (2011) Section 4.2, represent the average doctor-prescription amounts per consultation in cantons of the Midi-Pyrenees Department in southwest France. There are 268 cantons in the Midi-Pyrenees where average doctor-prescription amounts (per consultation) were reported for the period January 1, 1999-December 31, 1999. There are 32 "missing cantons" in the Midi-Pyrenees (MP) with no doctor-prescription data. Not all of the "missing cantons" in the files correspond to actual cantons; see below.

There are two datasets created from the original ones sent by Christine Thomas-Agnan: "Canton_neighbor.csv" and "Canton_vertex.csv". "Canton_neighbor.csv" is the dataset with cantons and their neighbours in the Midi-Pyrenees Region of France; note that only cantons with prescription data are included. "Canton_vertex.csv" is the dataset with cantons and vertices of the polygons; note that all cantons from the Midi-Pyrenees region are included, whether or not there were prescription data associated with them. The first 268 cantons in the dataset are those with prescription data. The same canton in both "Canton_neighbor.csv" and "Canton_vertex.csv" has the same value of "No.". The explanation of variables associated with each dataset is given as follows:

Format:

Canton_neighbor.csv

Variable Variable Explanation

No.: ID number from 1 to 268 (number of cantons with prescription data)

INSEE_ID: INSEE code

CANTON_NAME: Canton name

X: X-coordinate of the centroid (in meters according to NTF)

Y: Y-coordinate of the centroid (in meters according to NTF)

Z: Variable Z (average prescription amount per consultation in 1999, in French Francs)

X2: Variable X2 (percentage of patients 70 or older)

X1: Variable X1 (per-capita income)

E: Variable E (number of consultations in 1999)

N1-N12: INSEE_IDs of the neighbour cantons (up to 12 neighbors)

Canton_vertex.csv

Variable Variable Explanation

No.: ID number from 1 to 300 (number of all cantons in Midi-Pyrenees Region)

INSEE_ID: INSEE code

CANTON_NAME: Canton name

X: X-coordinate of the centroid (in meters according to NTF)

Y: Y-coordinate of the centroid (in meters according to NTF)

X1,Y1 - X76,Y76: (X,Y) coordinates of vertices of the corresponding CANTON (up to 76 vertices)

In addition to Cressie et al. (2005) and Section 4.2 of Cressie and Wikle (2011), these data have been analysed and modeled in Kang et al. (2009).

Canton_neighbor.csv: Doctor-Prescription Amounts per Consultation and other covariates in Midi-Pyrenees cantons. Also included are the cantons' neighbors. Note: only the 268 cantons with prescription data are included in this file.

Canton_vertex.csv: Verticies of the polygons describing all cantons, including "missing cantons"; the first 268 cantons in the dataset are those with doctor-prescription data.

Usage:

All Cantons: there are 32 "missing cantons" in the Midi-Pyrenees that do not have doctor-prescription data. Recall there are 268 cantons with doctor-prescription data. Then, there are 300 cantons in total.

Missing Cantons: the centroids of 32 "missing cantons"; since there are a total of 294 administrative cantons in the Midi-Pyrenees, 6 of the so-called missing cantons must be small polygons formed by digitization error. It is not known which of the 32 'cantons' whose centroids are shown in this file are real and which are artefacts of the digitization. This does not affect our analysis of the 268 cantons with data.

References:

N. Cressie, O. Perrin, and C. Thomas-Agnan. Likelihood-based estimation for Gaussian MRFs. Statistical Methodology, 2:1-16, 2005.

E. L. Kang, D. Liu, and N. Cressie. Statistical analysis of small-area data based on independence, spatial, non-hierarchical, and hierarchical models. Computational Statistics and Data Analysis, 53:3016-3032, 2009.

N. Cressie, and C. K. Wikle. Statistics for Spatio-temporal Data. John Wiley & Sons, 2011.