添加UP主视频数量分布分区多样性分布图

(* ::Package:: *)

(* \:4ece\:6570\:636e\:5e93\:83b7\:53d6UP\:4e3b\:5206\:533a\:591a\:6837\:6027\:6570\:636e *)

data = Normal[ExternalEvaluate[

  DatabaseReference[

   File[FileNameJoin[{NotebookDirectory[], "..", "data.db"}]]],

   "WITH up_stats AS (

    SELECT 

        author_id,

        COUNT(*) as video_count,

        COUNT(DISTINCT tidv2) as distinct_tidv2_count

    FROM videos

    WHERE tidv2 IS NOT NULL

    GROUP BY author_id

    HAVING COUNT(*) >= 10

)

SELECT 

    distinct_tidv2_count as x,

    COUNT(author_id) as y

FROM up_stats

GROUP BY distinct_tidv2_count

ORDER BY distinct_tidv2_count;"]];

(* \:8f6c\:6362\:4e3a\:5750\:6807\:70b9\:5217\:8868 *)

points = {#["x"], #["y"]} & /@ data;

(* \:521b\:5efa\:67f1\:72b6\:56fe *)

plot = BarChart[points[[All, 2]],

  ChartStyle -> RGBColor[0.368, 0.507, 0.71],

  Frame -> True,

  FrameLabel -> {"\:6d89\:730e\:7684\:5206\:533a\:6570\:91cf", "UP\:4e3b\:6570\:91cf"},

  FrameTicks -> {All, All},

  FrameStyle -> Directive[Black, 14],

  GridLines -> Automatic,

  GridLinesStyle -> Directive[Gray, Dotted],

  ImageSize -> 800,

  PlotLabel -> 

   Style["\:4e0a\:699c\:89c6\:9891\:6570\[GreaterEqual]10\:7684UP\:4e3b\:5206\:533a\:591a\:6837\:6027\:5206\:5e03", 16, Bold],

  LabelStyle -> Directive[Black, 12],

  Epilog -> {

    (* \:5728\:6bcf\:4e2a\:67f1\:5b50\:4e0a\:65b9\:663e\:793a\:6570\:91cf *)

    MapIndexed[

     Text[Style[#1, 10, Black], {First[#2], #1 + 

         Max[points[[All, 2]]]*0.01}] &, points[[All, 2]]],

    (* \:6dfb\:52a0\:7edf\:8ba1\:4fe1\:606f *)

    Text[

     Style[

      Row[{"\:603bUP\:4e3b\:6570: ", Total[points[[All, 2]]], 

        "  \:5e73\:5747\:6d89\:730e\:5206\:533a\:6570: ", 

        NumberForm[

         N[Total[points[[All, 1]]*points[[All, 2]]]/

          Total[points[[All, 2]]]], 3]}], 12, Black], 

     Scaled[{0.5, 0.95}]]

  }

];

(* \:663e\:793a\:56fe\:5f62 *)

plot

(* \:5bfc\:51fa\:4e3aPNG\:6587\:4ef6 *)

Export[FileNameJoin[{NotebookDirectory[], "up_diversity_distribution.png"}], plot]

(* \:67e5\:770b\:6d89\:730e\:5206\:533a\:6700\:591a\:7684UP\:4e3b *)

topDiverseUPs = Normal[ExternalEvaluate[

  DatabaseReference[

   File[FileNameJoin[{NotebookDirectory[], "..", "data.db"}]]],

   "WITH up_stats AS (

    SELECT 

        a.name as up_name,

        COUNT(v.id) as video_count,

        COUNT(DISTINCT v.tidv2) as distinct_tidv2_count

    FROM videos v

    JOIN authors a ON v.author_id = a.id

    WHERE v.tidv2 IS NOT NULL

    GROUP BY v.author_id

    HAVING COUNT(v.id) >= 10

)

SELECT 

    up_name,

    video_count,

    distinct_tidv2_count

FROM up_stats

ORDER BY distinct_tidv2_count DESC, video_count DESC

LIMIT 10;"]];

Print["\:6d89\:730e\:5206\:533a\:6700\:591a\:7684\:524d10\:540dUP\:4e3b\:ff1a"];

TableForm[{#["up_name"], #["video_count"],#["distinct_tidv2_count"]} & /@ topDiverseUPs, 

 TableHeadings -> {None, {"UP\:4e3b\:540d\:79f0", "\:4e0a\:699c\:89c6\:9891\:6570", "\:6d89\:730e\:5206\:533a\:6570"}}]