Machine learning projects become easier when you treat them as a workflow, not a magic trick.
Start by understanding the problem, inspecting the data, cleaning missing values, and creating features that actually describe the thing you want to predict.
After that, train a simple baseline model, measure it clearly, and improve one part at a time.