Data Preparation – ETL

Before diving into the Machine Learning or generally into the Analytics part, we need to prepare the data. Now, preparing the data can assume many different meanings: just shaping the dataset in the required form by the algorithm, blending it with another related dataset, handling missing values appropriately, generate new more meaningful features from the existing ones, reduce dimensionality by getting rid of uninformative features or on the opposite selecting the best descriptive features, removing outliers, reducing the dataset size via sampling or clustering, normalizing, and so on.

As you see, data preparation means much more than just shaping the dataset in the desired form. The next video shows a quick overview of the current most commonly used data processing techniques.

The workflows shown in this video can be found on the EXAMPLES server under 50_Applications/28_Predicting_Departure_Delays/01_Analytics50_Applications/28_Predicting_Departure_Delays/01_Analytics^*.