Meetings

Jump to bottom

Pritam Sukumar edited this page May 1, 2018 · 9 revisions

Meeting for roadmap (28/4/2018)

Trello "Problem Classification" board has assigned tasks now. 👍🏾

Overall integration plan (27/4/2018)

Whiteboard picture

Big picture

Metrics: need to define proper metrics to track from a business standpoint
Logging: need proper logging of to track metrics
Trends: need to be able to extract trends of relevant metrics/variables

Development platform

Research needed on:

What is available for ML on AWS
How python integrates, i.e. should we use their libraries? Our libraries, and so on..

Actual model

This is where we currently are at. Research needed on:

What algorithm to use:
- Currently working on Naive Bayes
- Next in line: Deep Learning
What features to use:
- Many variations possible of cleaning, embeddings, and tokenization. See the main Readme of this repo for some info
What accuracy measure to use
- Choices: overall accuracy, f-score (major and minor), precision, recall, ...More than one number is needed to get a proper picture of what the algorithm is doing
It would be helpful to get a proper system going to iterate over the parameters with least manual intervention