| 修订版 | 58bdda60df9632ec816ae58dd167966cd2869f88 (tree) |
|---|---|
| 时间 | 2024-08-30 17:13:08 |
| 作者 | Lorenzo Isella <lorenzo.isella@gmai...> |
| Commiter | Lorenzo Isella |
I added an automatic exploratory data analysis.
R-codes/clean_scoreboard.R (diff)| @@ -1,23 +1,28 @@ | ||
| 1 | 1 | rm(list=ls()) |
| 2 | -## last saved on Time-stamp: "2024-01-12 10:43:01 lorenzo" | |
| 2 | +## last saved on Time-stamp: "2024-08-30 10:09:30 lorenzo" | |
| 3 | 3 | |
| 4 | 4 | |
| 5 | 5 | library(tidyverse) |
| 6 | 6 | library(janitor) |
| 7 | 7 | library(openxlsx) |
| 8 | - | |
| 8 | +library(DataExplorer) | |
| 9 | 9 | |
| 10 | 10 | |
| 11 | 11 | source("/home/lorenzo/myprojects-hg/R-codes/stat_lib.R") |
| 12 | 12 | |
| 13 | -df<- read.xlsx("../input/scb_data_for_figures_2-22_top.xlsx") |> | |
| 14 | - as_tibble() |> | |
| 13 | +## df<- read.xlsx("../input/scb_data_for_figures-24-01-2024.xlsx") |> | |
| 14 | +## as_tibble() |> | |
| 15 | +## clean_names() | |
| 16 | +## ## clean_data() | |
| 17 | + | |
| 18 | +df<- read_csv("../input/scb_data_for_figures_LI.zip") |> | |
| 19 | + ## as_tibble() |> | |
| 15 | 20 | clean_names() |
| 16 | - ## clean_data() | |
| 17 | 21 | |
| 18 | -saveRDS(df, "scoreboard.RDS") | |
| 19 | 22 | |
| 20 | -write_tsv(df, "scoreboard.tsv.gz") | |
| 23 | +saveRDS(df, "../input/scoreboard.RDS") | |
| 24 | + | |
| 25 | +## write_tsv(df, "scoreboard.tsv.gz") | |
| 21 | 26 | |
| 22 | 27 | ## df_summary <- df %>% |
| 23 | 28 | ## group_by(year, member_state) %>% |
| @@ -29,4 +34,36 @@ | ||
| 29 | 34 | |
| 30 | 35 | ## saveRDS(df_summary, "scoreboard_aggregated_expenditure.RDS") |
| 31 | 36 | |
| 37 | + | |
| 38 | + | |
| 39 | + | |
| 40 | +config <- list( | |
| 41 | + "introduce" = list(), | |
| 42 | + "plot_intro" = list(), | |
| 43 | + "plot_str" = list( | |
| 44 | + "type" = "diagonal", | |
| 45 | + "fontSize" = 35, | |
| 46 | + "width" = 1000, | |
| 47 | + "margin" = list("left" = 350, "right" = 250) | |
| 48 | + ), | |
| 49 | + "plot_missing" = list(), | |
| 50 | + "plot_histogram" = list(), | |
| 51 | + "plot_density" = list(), | |
| 52 | + "plot_qq" = list(sampled_rows = 1000L), | |
| 53 | + "plot_bar" = list(), | |
| 54 | + "plot_correlation" = list("cor_args" = list("use" = "pairwise.complete.obs")), | |
| 55 | + "plot_prcomp" = list(), | |
| 56 | + "plot_boxplot" = list(), | |
| 57 | + "plot_scatterplot" = list(sampled_rows = 1000L) | |
| 58 | +) | |
| 59 | + | |
| 60 | + | |
| 61 | +create_report(df ,output_file = "scoreboard_report.html", | |
| 62 | + output_dir="../output" , | |
| 63 | + config = config | |
| 64 | + ) | |
| 65 | + | |
| 66 | + | |
| 67 | + | |
| 68 | + | |
| 32 | 69 | print("So far so good") |
Fork