Hello everyone.
I´d like to know the different models used to make predictions under differents circumstances. Are there any conventions established? By this I mean: Given a set of features X1 in regards to a certain topic T, which models have proved to do best in order to predict a target Y1?
In particular, I´ve been dealing with some data related to psychological traits and other about ML & AI user trends and am not quite sure which models to use in order to make accurate predictions. This is really something that prevents me from making observations and reaching conclusions.



------------------------------
[Marco] [Sánchez Sorondo]
[UBA]
------------------------------

#GlobalAIandDataScience
#GlobalDataScience

Hi @Marco Aurelio Sánchez Sorondo
​​​​
I am interested in this discussion and following the same.

My thinking is that different models fit different needs and the kind of  expected predictions, what you have referred to as "certain topic T". The features and outcomes could be linearly or non-linearly related. the target Y, could be continuous or could be discrete etc. What I am trying to say is that without these details, I guess it would be hard to give a general advice on what to use for your prediction model. A very  general direction though, would be to get to know your psychological traits and the user trends data... Can they be put into a classification case? in which you take on logistic regression models and if otherwise then take on linear regression models. But again if your data doesn't have labels then obviously you will need to consider other clustering models etc.

As to whether you are making accurate predictions, well, my take is that this is a question of testing the model chosen against the test data.

Disclaimer. Hope I understand your discussion well. I am yet to be an authority in this area and therefore this is a thought from my current level of understanding and open to further guidance. Please bear.

------------------------------
Michael Nganga.
Aspiring Data Scientist
------------------------------


Original Message:
Sent: Fri April 17, 2020 09:23 PM
From: Marco Aurelio Sánchez Sorondo
Subject: Model selection

Hello everyone.
I´d like to know the different models used to make predictions under differents circumstances. Are there any conventions established? By this I mean: Given a set of features X1 in regards to a certain topic T, which models have proved to do best in order to predict a target Y1?
In particular, I´ve been dealing with some data related to psychological traits and other about ML & AI user trends and am not quite sure which models to use in order to make accurate predictions. This is really something that prevents me from making observations and reaching conclusions.



------------------------------
[Marco] [Sánchez Sorondo]
[UBA]
------------------------------
#GlobalAIandDataScience
#GlobalDataScience

Thanks for your response Michael! Yes, of course... Different models fit different scenarios. I´ve been trying some models already, but maybe there is already a certain model that fits best a certain case. So you suggest to see first if the values are continuous or discrete?

------------------------------
[Marco] [Sánchez Sorondo]
[UBA]
------------------------------

Original Message:
Sent: Mon April 20, 2020 02:10 AM
From: Michael Nganga
Subject: Model selection

Hi @Marco Aurelio Sánchez Sorondo
​​​​
I am interested in this discussion and following the same.

My thinking is that different models fit different needs and the kind of  expected predictions, what you have referred to as "certain topic T". The features and outcomes could be linearly or non-linearly related. the target Y, could be continuous or could be discrete etc. What I am trying to say is that without these details, I guess it would be hard to give a general advice on what to use for your prediction model. A very  general direction though, would be to get to know your psychological traits and the user trends data... Can they be put into a classification case? in which you take on logistic regression models and if otherwise then take on linear regression models. But again if your data doesn't have labels then obviously you will need to consider other clustering models etc.

As to whether you are making accurate predictions, well, my take is that this is a question of testing the model chosen against the test data.

Disclaimer. Hope I understand your discussion well. I am yet to be an authority in this area and therefore this is a thought from my current level of understanding and open to further guidance. Please bear.

------------------------------
Michael Nganga.
Aspiring Data Scientist

Original Message:
Sent: Fri April 17, 2020 09:23 PM
From: Marco Aurelio Sánchez Sorondo
Subject: Model selection

Hello everyone.
I´d like to know the different models used to make predictions under differents circumstances. Are there any conventions established? By this I mean: Given a set of features X1 in regards to a certain topic T, which models have proved to do best in order to predict a target Y1?
In particular, I´ve been dealing with some data related to psychological traits and other about ML & AI user trends and am not quite sure which models to use in order to make accurate predictions. This is really something that prevents me from making observations and reaching conclusions.



------------------------------
[Marco] [Sánchez Sorondo]
[UBA]
------------------------------
#GlobalAIandDataScience
#GlobalDataScience

Original Message------

Hello everyone.
I´d like to know the different models used to make predictions under differents circumstances. Are there any conventions established? By this I mean: Given a set of features X1 in regards to a certain topic T, which models have proved to do best in order to predict a target Y1?
In particular, I´ve been dealing with some data related to psychological traits and other about ML & AI user trends and am not quite sure which models to use in order to make accurate predictions. This is really something that prevents me from making observations and reaching conclusions.



------------------------------
[Marco] [Sánchez Sorondo]
[UBA]
------------------------------

#GlobalAIandDataScience
#GlobalDataScience