I m trying to score "how much a product is expected in the market". I created some features: How much this product is used each year. Where was it used . how many product for each country. the main component of this product compared the product with other similar products. etc. I built a big matrix/table with all these information. I want to create a prediction model, where the input is a main component of a product and the output is ( expected with a pourcentage).

I dont have a training data ( just features without related results which indicates if the product is expected or no) I dont have historical data to confirm the results of my work.

Some advices ?

Thank you all

  • $\begingroup$ Predict the stock market tomorrow without any data from today or at any point in time? May as well generate a random number. Otherwise you could try clustering with many different products and their relevant information to get a sense of where this product might end up (in which cluster). $\endgroup$ Oct 9, 2018 at 8:23
  • $\begingroup$ The difficulty is that this product is innovation so no similar products are done before and no labels .. $\endgroup$
    – Lizou
    Oct 9, 2018 at 8:28
  • $\begingroup$ Then there is no real way to tell. Maybe the guys in marketing can use their usual tricks to get an estimate, for ex. conduct a poll asking people how likely you are to use this product and such. $\endgroup$ Oct 9, 2018 at 8:34

1 Answer 1


If you do not have data then I think the problem is more of research than Machine Learning. Ask your research team to gather primary data about the product by conducting surveys, polls,etc.

Depending upon the data you get from research, you can use some statistical methods like hypothesis testing to solve the problem.

  • $\begingroup$ Thank you for the answer, I think some statistics to obtain training data is not a bad idea as a staring point. I cant do surveys because the data is international... $\endgroup$
    – Lizou
    Oct 9, 2018 at 14:46
  • $\begingroup$ Another option if you can obtain some statistics could be Monte Carlo simulation. $\endgroup$
    – Skiddles
    Nov 8, 2018 at 13:49

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge that you have read and understand our privacy policy and code of conduct.