XY Titel

Benchmarking Feature Selection Algorithms with Xy()

André Bleier Blog, Data Science

Feature Selection Feature Selection is one of the most interesting fields in machine learning in my opinion. It is a boundary point of two different perspectives on machine learning – performance and inference. From a performance point of view, feature selection is typically used to increase the model performance or to reduce the complexity of the problem in order to …

Food for Regression: Using Sales Data to Identify Price Elasticity

Daniel Lüttgau Blog, Data Science

A few hundred meters from our office, there is a little lunch place. It is part of a small chain that specializes in assemble-yourself, ready-to-eat salads. When we moved into our new office a few years ago, this salad vendor quickly became a daily fixture. However, overtime, this changed. We still eat there regularly, but I am certain, if one …

XY Titel

Pushing Ordinary Least Squares to the limit with Xy()

André Bleier Blog, Data Science

Introduction to Xy() Simulation is mostly about answering particular research questions. Whenever the word simulation appears somewhere in a discussion, everyone knows that this means additional effort. At STATWORX we are using simulations as a first step to proof concepts we are developing. Sometimes such a simulation is simple, in other cases a simulation is plenty of work. Though, research …

pandas vs. data.table – A study of data-frames

Christian Moreau Blog, Data Science

Overview and Setting Python and R have become the most important languages in analytics and data science. Usually a data scientist can at least navigate one language with relative ease and at STATWORX we luckily have both expertises available. While, with enough will and effort, any coding project can be completed in either language, perhaps they differ in some perfomance …

XY Titel

Simulating Regression Data with Xy

André Bleier Blog, Data Science, Statistik

In a recent project, I have developed a gradient boosting algorithm to estimate price elasticities. Surely, it is necessary to validate if the functionalities of the algorithm are working as intended. I started using nonlinear time series data from another blog post about lag selection as a validation basis. Unfortunately, at that time I did not wrap the simulation code …

Sparse Matrizen – wann sollte man sie nutzen?

Jakob Gepp Blog, Data Science, Statistik

Wenn man mit Matrizen arbeitet, die viele Nullen enthalten, dann sind schwachbesetzte (engl. sparse) Matrizen das richtige. Hierbei wird der benötigte Speicherplatz der Matrix reduziert, in dem der Inhalt der Matrix effizienter verwaltet wird. Es gibt verschiedene Methoden Matrizen zu komprimieren – zum Beispiel in dem nur die Tupel aus Zeile, Spalte und Wert genutzt werden. Die Matrix reduziert sich …

Gradient Boosting – Identifikation nichtlinearer autokorrelativer Einflüsse

André Bleier Blog, Data Science

Autokorrelation ist ein zentrales Element der Zeitreihenanalyse. Um den Nutzen greifen zu können, sollten wir zunächst einmal klären, was Autokorrelation überhaupt ist. Autokorrelation Korrelation beschreibt Einflüsse bzw. Abhängigkeiten von Zufallszahlen. Eine Autokorrelation hingegen beschreibt den Einfluss einer Zufallszahl auf vergangene Zeitpunkte. Das bedeutet, man vermutet bei solch einer Untersuchung, dass bei einer Zufallszahl zeitliche Abhängigkeiten bzw. Einflüsse vorzufinden sind. Autokorrelation …