1. A computer-implemented method to identify irregularities in the distribution of electronic files within provider networks, wherein each provider network comprises at least one server computer which is adapted to distribute the electronic files to a multitude of client computers within the provider network, wherein the method includes the following steps:a. defining, by a selection unit in a data analysis device, a number N of discrete time periods t1 . . . tN,
b. selecting, by the selection unit, out of the electronic files a specific query file,
c. selecting, by the selection unit, a set of D provider networks, wherein the number D is equal or larger than three,
d. acquiring, by an acquisition unit in the data analysis device, from the D server computers associated with the D selected provider networks, a set of time series data streams p(t)=p1(t), p2(t), . . . , pD(t), which are indicative of a common metric that has been associated to the query file,
e. performing, by a transformation unit in the data analysis device, an isometric logratio transformation on p(t) to compute a compositional data set z(t)=z1(t), z2(t), . . . , zD(t),
f. filtering, by a filtering unit in the data analysis device, the values of z(t)=z1(t), z2(t), . . . , zD(t) using a local robust filtering model to compute local estimates ?(t)=(t), (t) . . . , (t), wherein repeated median regression is applied by the filtering unit and the choice of the regression window width is based on a test of the slopes of the regression line within the window using the SCARM (Slope Comparing Adaptive Repeated Median) method,
g. computing, by an outlier detection unit in the data analysis device, the residuals rl(t)=zl(t)??l(t) for I=1, 2, . . . , D, and performing a multivariate outlier detection on rl(t) to calculate an outlier metric, and
h. responsive to the calculated outlier metric, computing, by an evaluation and reporting unit in the data analysis device, a set of control data to control the distribution of electronic files in the provider networks, and sending the control data to the server computers, wherein the control data causes the server computers to amend the distribution of electronic files.