Predict statistically sales trend

Question

I am currently doing statistical software that supports POS software. They asked me to predict how this month's sale will be based on the same month but of previous years (5 years ago).

Example: Predict the sales trend for the month of March of this year. I currently have sales from March of the years 2013 to 2017, and based on that, I must predict how we should go in March of this year. The problem is that I do not know what to do, I have investigated linear regression, or trends but I still do not understand.

I think that I should predict day by day the sale of each of the days, each day would show as a scatter chart and from there I could extrapolate one more point to get the trend, but I have no idea either.

Any help will be welcome. Thanks.

EDIT 1 I really have all the sales every day for 5 years, but the client tells me that I have to predict what the sale will be like this month of - for example - March, and I already have all that data of all the sales, and I would like that, from the sale of all the days of March of all the 5 previous years, to predict how it will be the sale of this March of this year.

Answer 1

https://stats.stackexchange.com/search?q=user%3A3382+daily+sales+data will give you a number of discussions regarding predicting daily data and subsequentally month end totals . The issue is you need to integrate memory i.e ARIMA structure along with deterministic factors like day-of-the-week; week-of-the-month ; month-of-the-tear; day-of-the-month , level shifts ; local time trends ; lead and lag effects around the holidays ; long-weekend effects et al. Post your data in a csv file and also indicate starting date and country.