Records from ledgers of Dutch companies all across the Netherlands are used in this study. Records can be submitted in the ledgers with various lags, because the data of many different bookkeepers is involved with different workflows. Bookkeepers can be punctual or late, therefore records can be submitted with various lags in the ledgers. This causes missing data, which results in a deformation of a time series that is constructed from these records. Using a technique called now­casting, a prediction can be made of how these series with no missing data would have looked like.
This study sheds light on how information of an incomplete time series from ledgers of Dutch companies can be used to now­cast on that series, without the use of external indicators. To better utilize the information available from the series, an addition to the Seasonal Auto Regressive Integrated Moving Average with eXogenous regressors (SARIMAX) model is proposed. The addition to the SARIMAX model is presented in two forms: the additive­ and multiplicative relation between indicator­ and target series. These are modeled with the goal to improve the information utilization and therefore improve the now­casting accuracy. Experiments have shown that this addition to the model does not give a direct improvement in accuracy compared to an ordinary SARIMAX model. Thereafter an iterative now­cast procedure is proposed to utilize information from highly lagged records. It has been shown that this gives a slight increase in accuracy for the overall now­cast.

This report describes the Bachelor End Project, a compulsory course for the Bachelor Computer Science at Delft University of Technology. This course has a group of 4 students complete a project for a real-world company (the client). The client for this project was Relive, a company that has a tracker for outdoor activities with a focus on the experience of the activity. They wanted a route recommendation engine for their users, in particular users that are unfamiliar with the area. The specific task from Relive was to use the Relive database to find popular, dissimilar routes to recommend within an area, and then find context for those recommended routes. The first attempt to find these routes was using a heatmap to see which routes were popular, and then clustering routes to find dissimilar routes to recommend. This had some problems with unwanted biases towards cities, so the second attempt switched it around by first clustering on general area to make sure recommendations are dissimilar, and then using clustering and finding big clusters to see which routes are popular. This last approach worked decently: it can recommend 73% of the expected routes. One of the key problems for now is the long waiting times. There were two kinds of context that were considered: keywords and photos. Keywords can be extracted by processing titles from nearby routes with significant search. Photos are added to routes by the users but may not be relevant as context. Examples of such useless photos are selfies, which also bring privacy concerns. Due to time constraints, giving context was not integrated into the final product. The final product is a web application client and a server for that client. The users interact with the web application, and the recommendation algorithm is run on the server.