Southern California Statistical American Statistical Association #ASA#SCASA is having an upcoming all day workshop for Causal Inferences by Brian Vegetabile. #rstats#Statistics For more information please head over to SCASA website (https://t.co/gZg9TTHQo2).
@mpjdem @rstatstweet I get accuracy between 70% to 79% on same test data set when you build model with same seed number but each model have different column order. I'm sure that different seed with same column orders will produce different fit model since Random Forest bootstrap obs & predictors.
TIL: R's Random Forest package is highly dependent on the order of the predictors in your dataframe. Two RF models with the same data but different order of columns/predictors in the dataframe will be two different models. #rstats#Statistics#MachineLearning
@Lee__Drake@rstatstweet Correct. You normally shouldn't shuffle your data via columns unless you're writing a thesis and you're algorithm requires it. 😭 It also make sense why predictors' order matter because at every tree split, Random Forest bootstrap a subset of predictors.
I got to meet @f2harrell Learned a lot about regression modeling. Made me feel bad about my modeling strategies for my thesis. His philosophy and take on modeling will help me to be a better statistician. It feels like standing on the shoulders of giants. #statistic
@joeerl You've contributed to who I am as a programmer as of today. Your talks and blog posts are the reason why I am better. The energy, passion, and kindness you give in those talks and blog posts is something I will dearly miss. You're my inspiration. RIP. #rememberingjoe
I've been updating the Southern California Chapter of the American Statistical Association (SCASA) webpages, got an awesome workshop coming up. Regression Modeling Strategies by Dr. Frank Harrell! https://t.co/gZg9TTHQo2 #statistics#regressionmodels#datascience#machinelearning
My raw data messed up on gender. So I wrote a code that read each item's description and count number of time the word women and men are mentioned to determine the product intended gender. I shall call this NLP. #NLP#Datascience#machinelearning#artificialintelligence
I didn't realize how slow it was to copy paste 1.4 MB of text worth of SQL to terminal. Should have pipe it to psql via script. Webscrape, Data modeling, and ETL is pretty dang fun. I got ~500 MB worth of raw text data left for ETL. T____T #sql#scripting#terminal
I'm now the web master for American Statistical Association for Southern California. I added the December 2018 newsletter for Southern California Chapter of the American Statistical Association (SCASA). Check it out at https://t.co/qA5i1Ie7WO #newsletters#statistics
I'm doing ARIMA Time Series forecasting in R presentation at LA East R User Group (Jan 14, 2019). https://t.co/wQNmwth9Ix #rstats#statistics#forecasting#arima