Model Contract & Research

[selvarajuv]

DUE 09/22 11:59 PM

Identify potential data contracts for our model
Identify potential models we can use

What we mean by "data contract":

• What is the shape of the data that the model expects?
• What kind of feature engineering will need to be done for the most accurate and efficient modeling?
• Are there any other factors that might be necessary or beneficial for an accurate model?
• What does the model output look like?
• What will be useful to the client?

Model considerations:

• What are we trying to answer?
• What kind of data do we have available to us?
• How much time is required to train/evaluate?
• How can we continue to refine the model based on user feedback.

We believe that training something a Neural Network could be extremely interesting. Look into that option and see if it viable.

Acceptance criteria:

• Documentation on what the model contract will look like (inputs vs outputs)
• Documentation of at least three models we can use and their pros and cons (have one of them be a fully self trained neural network)
• Be ready to present findings next meeting (09/23)