What is obvious to you is NOT obvious to others. - Share your knowledge - Ask for help when you are stuck

The internet provides access to an education that the aristocracy of old couldn't have imagined. It also provides the perfect attack vector for marketers to exploit cognitive vulnerabilities and dominate your attention. A world-class education is free for the undistractable.

sorted is a function and sort is an attribute of the object. #python

A few folks are asking based on this tweet if I've given up on Jupyter... No! Jupyter notebooks remain the best option for exploratory data analysis, reproducible documents, sharing of results, tutorials, etc. For software & package development, I've always used other tools.

"scikit" is not an acceptable shorthand for "scikit-learn".

parse_number is one of my favorite functions (file under "how to avoid regex"), just watch out for the negatives: > parse_number(c("wk 3", "week-1", "7", "w#9")) [1] 3 -1 7 9 > abs(parse_number(c("wk 3", "week-1", "7", "w#9"))) [1] 3 1 7 9 #readr #tidyverse #rstats

Replacing missing values with mean is a statistical malpractice. Take into account the domain knowledge, presence of outliers and the shape of distribution your data takes.

keras package is published by Rstudio and kerasR by the R community. If you are comfortable with tidyverse, you should go with the keras package. Installation of the former is easier than the latter. You need to setup python env if you go with the kerasR package. #rstats

Always take special care if your data has 0 values in any of the columns. If there are zero values, you may not be able to successfully preprocess your data. Log transformation does not work on zero values.

#R val <<- val + 1 The '<<-' is a superassignment operator. It is used to modify variables in the global environment. Not recommended to use as it can mess up assignments. Use can be in writing of sequence generators. #rstats

New to (Artificial) Neural Network? Take a look at karpathy.github.io/neuralnet… Karpathy explains the basics using circuit gates and simplifies the intuition behind the concepts. The blog is long and can be boring but I'd recommend to complete it. I enjoyed it. #neuralnets

#R The operator %<>% (package - magrittr) will not only perform operations on the RHS but also assign the changes on the object at LHS. #rstats

#R read_csv() is better than read.csv(). Not only it is faster but also recognizes format and does the conversion. I have a date time col which was imported as factor in read.csv but read_csv converted it to the desired POSIXct format. #rstats