I would like to have some ideas about each customer depending on how many products he sells each month.
We have created a new feature X based on some other features.
We have noticed that, in most case, when Feature X decrease the amount of product sells decrease as it is illustrated with the figure bellow
For each customer, I have a similar plot (i.e in the most cases when the feature decrease buying product decrease).
How can I generalize this analysis for getting one global plot?
I think your data has three variables: customer_ID, number_of_products_sold, and feature_x. Time is not, as far as I can tell from your description, relevant.
To get all three relevant variables plotted in one figure, start with the data for a single customer and show number_of_products_sold versus feature_x as a scatter plot (removing the time aspect). This scatter plot with, say, circles for each data point, could be extended to multiple customer_IDs by overlaying scatter plots with different marker types.
