Finding or making large datasets is already a challenge in itself; but if they are completely unlabelled, they may end up being of little value as is. As such, data labelling marks a true cornerstone of the machine learning model building process.
Since models often need big data for precise results, there may well be millions of data points to process and label. Understandably, because of the need for quality in such a high volume of data, the labelling process ends up being one of the most time-consuming tasks in data science.
This guide aims to ease the burden of labelling time-series data by suggesting relevant approaches to the process and uncovering the most common pitfalls.