Data Science
Machine learning & data science for beginners and experts alike.
Beta, Pearson, Spearman, macros and more — last month’s Data Science Portal discussions covered a lot of territory! Here are three interesting conversations that happened in August. (And it’s not too late to revisit July’s highlights, if you missed them!)
Calculating Beta with Alteryx
@Joker_Hazard needed a method to calculate beta using Alteryx, based on covariance and variance. This metric is often used to assess a stock’s volatility relative to the entire market. @atcodedog05 and @apathetichell jumped in to help with tips, plus a workflow that can easily handle the 1,500 companies @Joker_Hazard wants to check out.
Better calculate beta. Image via GIPHY
Updating Parameters for New Data
Two interesting conversations revolved around how to make a tool more adaptable for a specific analysis. First, @Rohit_Karvekar asked how to adjust the preset minimum value for support in the MB Rules Tool. @danilang and @apathetichell offered suggestions for altering the R code underlying the tool for this purpose.
More recently, @neilgallen, @phottovy, @danilang and @mceleavey have been discussing how to dynamically adjust the parameters of the Find Nearest Neighbors Tool. This challenge has yet to be solved, so hop in if you have ideas!
Yes, that is who you think it is in blue. Image via GIPHY
These questions are a great reminder that there’s more to the R-based predictive tools than meets the eye. You can tinker with the code within them, and also obtain the models they generate to use elsewhere.
The Perfect Correlation
Finally, @suresh_abeyweera asked a great question about using the Pearson Correlation Tool as part of developing a recommendation engine, and wanting to group data by users to calculate correlations among ratings. @phottovy pointed out that the Spearman Correlation Tool has a built-in group by option, but the Pearson tool doesn’t — then went the extra mile and shared a batch macro that groups and then calculates the Pearson correlations. Amazing! (Also, if you’d like a refresher on Pearson vs. Spearman correlations, here’s a useful guide.)
Thanks to everyone who participated in these great discussions! Also, be sure to stop by the Data Science Mixer podcast pages for our Cocktail Conversations. In August, we talked on the podcast about the role of data science in the grocery supply chain and in human resources — if you eat food and/or work, I promise you’ll find these episodes compelling. Cheers!
Image via GIPHY
Blog teaser photo by Mike Kononov on Unsplash
