In f0nzie/petroleum.ml: Machine Learning Applications in Machine Learning

Read Excel file with Petroleum Engineering applications

The Excel file has been manually modified by the user to enter which petroleum engineering discipline the paper belongs to. This is a very tiring activity since the user has to go paper by paper and find whichdiscipline falls into.

# read the user modified Excel file with PE discipline assignment
devres_loc <- "./inst/results"
xls.nodups.papers.4 <- readxl::read_xlsx(paste(devres_loc, "ml_nodups_papers_4.xlsx", sep = "/"), sheet = 1)
xls.nodups.papers.4

# unique PE applications
unique(xls.nodups.papers.4$pe_app)

library(dplyr)

# group by application and algorithm
grouped.5 <- xls.nodups.papers.4 %>% 
    select(-c(X__1)) %>% 
    group_by(pe_app, V3) %>% 
    summarize(avg_count = n()) %>% 
    rename(algorithm = V3) %>% 
    arrange(desc(algorithm)) %>% 
    print

Note. There are plenty of rows that are NA because we didn't go through all the papers to assign Petroleum Engineering application (pe_app).

An alternative here is applying some machine learning algorithm to classify the papers following the training data (those papers that were assigned a pe_app). But that would be misleading since we really don't know if a paper really used a specific ML algorithm unless we open and analyze the paper itself; the classification that has been performed based on the paper's title is flawd but at least gives us an idea of what we are pursuing.

The best route is doing samplng on the non-duplicate papers.

f0nzie/petroleum.ml documentation built on May 12, 2019, 6:22 p.m.