Data Science lifecycle be automated
In recent months, many tools have appeared that claim to automate all or parts of the data science process. How do they work? Could you build one yourself? If you adopt one of these tools, how much work would be necessary to adapt it to your own problem and your own set of data? Usually, the price to pay for automated machine learning is the loss of control to a black box kind of model. What you gain in automation, you lose in fine-tuning or interpretability. Can the entire data science lifecycle be automated? Can a machine learning model be created automatically from a set of data? In fact, a lot of solutions have come out recently that say they can automate all or parts of the data science process. How do they function? Could you construct one on your own? How much effort would it take to customize one of these tools for your unique situation and set of data if you adopted it? The loss of control to a black box model is typically the cost of automated machine learning. Automation comes at the cost of fine-tuning or interpretability workflow was designed for business analysts to easily create predictive analytics solutions by applying their domain knowledge. In this article, we will show the steps of this application from the business analyst point of view, when running from a web browser. In a follow-up article, we will show the behind-the-scenes implementation, explaining in detail the techniques used for feature engineering, machine learning, outlier detection, feature selection, parameter optimization, and model evaluation.
.jpg)
