The task is to predict the quarterly revenue numbers using machine learning. Only 28 quarterly data points for financial numbers are available as companies release the revenue data quarterly. I have few questions regarding this -
- Is it worth to apply machine learning models to these 28 points to predict the sales? I am skeptical that it would give good results because it seems the data points are too less for machine learning algorithms.
- How should I deal with the COVID years data as it completely distorts the trend of the series?
- This is the
seasonal_decomposeof the 28 data points. By looking at the plot, I am certain that there is a trend. But would it be advisable to apply time series on such small dataset?
One idea is to take the revenue for entire quarter divide by 3 to get avg. monthly revenue and attach to each month. Example the revenue is 900 million; give each of the three month in quarter 300 million that would increase the data points from 28 to 84.