Lately, several companies in the waste collection market installed sensors on the collection vehicles, which are then able to measure the waste weight present in the collected dumpsters. By using this data, with geostatistical-based analysis, it is possible to estimate the waste production in the served area. The objective of the thesis is to develop and implement statistical models that can be used to describe the territory waste production. In particular, in statistical applications and in econometrics it is common use to rely on data panels, which are multidimensional data involving measurements over time, and on spatial autoregressive models (SAR), to represent these random processes. A challenge that rises here is how to handle the unevenly spaced observations, the missing data and the high computational complexity.

Keywords: VAR, time series, statistics, high dimensionality