Making an R package with usethis, and UCI Capstone

June 25, 2019

Details This month we have Emil Hvitfeldt coming to talk about making an R package with usethis. Building your code into R packages is simple and allows you to easily manage and reuse your code.

The Paul Merage School of Business at UCI is graduating their second cohort of students from their MS in Business Analytics program. Last year we invited them to present their Capstone projects and it was a huge success, so they are back this year. We will have two groups presenting. One worked with the Amazon Alexa Skill Store and one worked with Pacific Life. Check out their abstracts for more details.

Come early and network. Do not forget to purchase a raffle ticket or two. It helps support the meet-up and we have a great prize. A book of your choice from CRC Press. Yup, you get to chose the book!

== Schedule == 6:30 - 6:50 Networking 6:50 - 7:00 Welcome & general announcement 7:00 - 7:15 UCI - Amazon Alexa Skill Store 7:15 - 7:30 UCI - Pacific Life 7:30 - 8:00 Making an R package with usethis 8:00 - Raffle 8:00 - 8:30 Networking and Clean-up

== UCI MSBA Capstone Talks ==

Pacific Life: Predicting policyholder lapse behavior

Applying the principles of Statistics and Machine Learning to predict policyholder lapse rate for Variable Annuities. In addition, we gained insight into the underlying dynamics of policyholder lapse behavior.

Amazon Alexa Skill Store:

Using effectively the skill intent model (a description of a third party application), including but not limited to, description, sample utterances, keywords, maturity rating, etc. we try to classify Alexa skills into one of 20 categories. Creating features with text data has a two-fold approach. The first is to use frequency of words to give words to each word occurring in the skill-intent model and use this matrix to predict the category as a supervised classification problem. The second is to use the semantic meaning, that is similarity of words, defined by cosine similarity. To overcome the big challenge of overlapping categories, we use a probabilistic approach to categorization.

== Making an R package with usethis ==

Speaker: Emil Hvitfeldt

Building an R package can seem daunting with its many files and structure. This talk will go through the different use cases for an R package, does and don’ts and best practices.

comments powered by Disqus