I have a huge array of products and their prices timeseries data. I want to find the correlation between those products prices. Most of the data has null values when the product is not yet into the market. What should I fill null values in each column with? I can use mean or median but that would be like correlating using wrong data. Filling with 0 or next first non null value (fillna='bfill') in the same column are any other better choice. What would you suggest? Greatly appreciated your responses. Continue reading...
