This dissertation investigates time-series analysis on Telecommunication Traffic data. The data set is comprised of two time-series of 4G-LTE technology, where one represents the Downlink and the other the Uplink speed for one Base Station. The time series has an hourly frequency and the training set has a total of thirty-seven (37) days and the test set up to sixteen (16) days. We initially perform a Visual and Statistical exploration to gain insight for applying the according algorithms. Data preprocess was necessary, since we have real-life data, where missing and erroneous values occur. Also, a look into statistical and econometric approaches takes place, in order to find the optimal algorithms with respect to the special characteristics of the given data. Briefly, the dissertation describes the procedure for using Python as the mean to efficiently and accurately provide forecasts on a time series to gain knowledge for effective management and planning for a Telecommunication network. Lastly, the data set used was for the months February to April of 2014, in Greece.
Collections
Show Collections
