Date

Attendees

Web Group
Jon demonstrated his latest web development.
http://www.jwoolfden.com/TSSG/multipage/index.html

Andy L. likes the original, vertical stacking for the schedule.
Discussion about using text in footer icons.
Further changes requested from Jon:
1. put schedule back to vertical stack view
2. use footer icon version where the icons include text
3. color of header and footer should match
4. solve all problems related to cut-off of content on mobile view
5. remove footer when mobile view. Hide with bootstrap based on screen size. Low priority request.
6. lighten the link items on the navbar so that they are readable. Closer to what we have now in tssg.tech.

Discussion on replying to requests/questions from possible new members to TSSG. 
Paul showed his version of the new web site and showed the code that includes Ajax commands. Presents smoothly on all browsers and phones.
Jon stored his code in the multipage branch of the tssgTechMVP.git repository on the server.

Mobile Group
Mark P. had to leave the meeting early: Nothing formal to present. Was going to talk about google assistant and maybe demo his test application.  It is related to mobile since assistant is available for ios and android and there is a mobile sdk.

Peter showed what he was working on related to the EventBoss2 project on Mobile.

Data Analytics Group
Ralph - working on getting a new GPU system set up. Should be especially useful for neural network testing.

Andy showed his work that he's doing with Mike. Insurance company data project from Kaggle.

Some notes from the presentation that I scribbled down...

Has a .csv file of data to clean.
Task of preparing data
nominal types of data - no order; just categories
ordinal values - t-shirt size s/m/l 1/2/3 has an order
binary values - 0/1 male/female
interval,  numerical data - height 
-1 means it's missing data

training data with target values
test data has no target values and we want to predict them.
Need to work with data that tells us something.
under sampling - remove data where value is = 0.
Need to do (PCA) principal component analysis - find out how data effects result
modify the test data with the training data.
data preparation and feature engineering = 80% of the work
Anyone have experience with the Eigenvalues problem and linear algebra?