I have summarised my own systematic approach towards a general Machine Learning problem that I follow for analysing and to be able to successfully approximate a really efficient model to make predictions with decently high accuracy.
It is a general issue when people run into some constraints while cleaning data or even analysing them and get stuck at a point where they can't figure out what to do next to make it right. The way I have curated the notebook help to mitigate this issue quite successfully and in a really clean way.
This approach is not the most efficient approach but following one keeps the developer in control rather than the data controlling the developer by being all over the place, where you just keep running around to make it right at every corner but end up messing it up more.