Hi,

Recently, I had to work on a simple time-series analysis. I performed poorly since I never worked with time-series before. I believe in a deterministic world, and in general, I prefer to find the causality of a specific data behavior prior to a simple way of empiristic modeling. However, I understand the need for time-series analysis as not enough data available, the underlying processes understood, the complexity bearable, or the time/need for a proper process understanding. The goal is to make a prediction based on the previously observed observations. In a traditional sense (Arima), you look at the trend, seasonality, and cycles - in the more modern way, you throw the data into a model architecture (deep learning). In this context, I should mention the famous paper Statistical Modeling: The Two Cultures, while I prefer to use algorithmic models and treat the data mechanism as unknown. I would add that the underlying data mechanism is deterministic, and we should use collected data to get improved models. Anyway, let’s use the many resources in the time-series field to get better in this field.

As a first step, I set my goal to get a robust understanding of time series and get some experience with different approaches. Perfection takes time, which is not my goal. At least, I want to be prepared for possible tasks. Therefore, I accomplished two MOOCS about time series. One is Practical Time Series Analysis to get a traditional view on the field, and the other one is Sequences, Time Series and Prediction as a deep learning approach.

Since university times, I have followed lectures by getting some reading material. This habit hasn’t changed after my transition into tech. I got the following books for theoretical and practical fundament:

The benefits are clear. If I don’t understand something correctly, I can look after it and research it. Also, with multiple sources, I get different views on the same topic - everyone has their own way of teaching. I see the experience of multiple experts in the field, which improves my learning and understanding.

Finally, for practical experience, I trained on given toy examples from both MOOCs. I translated them from R in Python or used a different framework (for example: using Pytorch instead of Tensorflow), Here are my hastily composed notebooks:

The initial learning took me around ten days, and it’s clear that I’m not an expert. But this is a first step to get familiar with the topic. Until now, I’m not even sure if I will need a time-series analysis in the future. Also, one of the best ways of learning is to get your hands dirty via writing and coding. From my perspective, I have some robust fundamentals to get deeper into this field if needed.

Thank you for your attention.