1. The nature of the sample data

2. the purpose of the project (classification, detection etc)

I want also add such criterion as complexity and speed of learning algorythm

Hi,

few points I shall consider:

1. The data and their distribution. We know some algorithms are more tolerant than others.

2. Which kind of transformation I could perform if possible  

3. Type of problem supervise or unsupervised 

4. Which kind of infrastructure do I have or can I have access to AWS for example.  This is more relevant to the tuning of the model and the processing time for example if you have 300 variables let say on half million observations and you tune the mtry with multiple folds using  random forest as example... well you time for a cup of coffee

5. The last one can I conduct the final optimisation. Some algorithms demand good knowledge to achieve the best tuning. 

I am sure there are other criteria and then when you have all you pick up a book such as Applied Predictive Modelling and review each algorithms with you answers. 

Hope this help.

Alain

Thank You all,

Michel Thanks you for such a wonderful material which gives great information.

 Alain i agree to you points . Thank you for your patience in answer this question

Abass and Vsevolod..thank you for your simple answers