machine-learning-split-80-20-l-regression

In this analysis we investigate what factors affect or drive the most, revenue (y). We can analyse past marketing campaigns and suggest which one works better, suggest changes and optimisation, and suggest how much money to invest or is possible to produce. we only feed the model with part of the data and test if the model can predict the rest. E.g. we only put 80pct of the data (training data), algorithm learns from this 80 pct and should able to predict the 20pct left, which is unseen data for the model. This is why we split the data (train/test). At evaluation, we provide the other 20pct left (not including the y variable, although we have the actual y variable) and compare the model output to the actual y to see how good the model is)