All services have two opponent forces fighting each other — user acquisition and user churn. As long as you acquire more than you lose, you are on a good road to success. If you’re not, there is some things you can do:
Let us disregard acquiring more users for now, as it is expensive and will only become more expensive as time goes by.
That leaves us with reducing churn. In a data driven company, luckily we know quite a lot about our users, and if we are lucky (and we are in…
According to the StackOverflow 2017 survey - how well can you predict your salary and what does it tell you about how to increase this in the future (from a data guys point of view).
For this analysis I used Python in Jupyter Notebooks to predict prices with a linear model. The code can be found on Github.
Let’s start with finding a good linear model to predict salary. For this I wrote a grid search that iterates over different fractions of the entire data set provided for training the model to counter overfitting.