Update results.py

results.py

@@ -125,47 +125,42 @@ fig.update_layout(
 
 Perplexity_Across_Different_Buckets_global_graph = fig
 
-import plotly.graph_objects as go
-
-# The data you provided
-DATA = [
-    ["2014", [["1-1", "2-5", "6-10", "11-100", "101-1000", "1001-30000000"], [17.410227605477868, 16.11176217183986, 15.632757662414805, 15.446116676532212, 16.716943171826703, 18.156821563322765]]],
-    ["2015", [["1-1", "2-5", "6-10", "11-100", "101-1000", "1001-30000000"], [17.446573602753478, 16.14852530113782, 15.627408549576069, 15.0055028132117, 15.565430373421485, 17.314701050452452]]],
-    ["2016", [["1-1", "2-5", "6-10", "11-100", "101-1000", "1001-30000000"], [17.307221780905284, 16.297702171159543, 15.948641884223639, 14.799690714225637, 14.935989931859659, 16.09585768919658]]],
-    ["2017", [["1-1", "2-5", "6-10", "11-100", "101-1000", "1001-30000000"], [17.338525603992114, 15.960924352297502, 15.912187993988933, 14.822102470001267, 14.778913482337416, 15.428145290012955]]],
-    ["2018", [["1-1", "2-5", "6-10", "11-100", "101-1000", "1001-30000000"], [17.08551151136689, 16.187802102106698, 14.935072408852303, 14.832038213200583, 14.508674264491997, 14.800605964649103]]],
-    ["2019", [["1-1", "2-5", "6-10", "11-100", "101-1000", "1001-30000000"], [16.818363305107052, 16.474269837858706, 14.944741674400241, 14.568394784374943, 14.690158822673334, 15.990949424635108]]],
-    ["2020", [["1-1", "2-5", "6-10", "11-100", "101-1000", "1001-30000000"], [16.98821894111693, 15.936494557783181, 14.79960386342691, 14.435682562274105, 14.58651834886038, 15.869365567783806]]],
-    ["2021", [["1-1", "2-5", "6-10", "11-100", "101-1000", "1001-30000000"], [17.125795647512877, 15.780419457145868, 14.631430892394002, 14.276477514399625, 14.337146941773641, 15.872474774329305]]],
-    ["2022", [["1-1", "2-5", "6-10", "11-100", "101-1000", "1001-30000000"], [16.573462144306383, 15.283018703313582, 14.378277745163881, 14.0611924390084, 13.9886330091318, 15.769421394877273]]],
-    ["2023", [["1-1", "2-5", "6-10", "11-100", "101-1000", "1001-30000000"], [15.4293630385597, 14.608379914730168, 14.118271697056592, 13.880215644749589, 13.767106666731275, 15.05749135510839]]]
-]
+##graph 2
 
-# Extract years and ranges (buckets)
-years = [year_data[0] for year_data in DATA]
-ranges = DATA[0][1][0]
-all_values = [year_data[1][1] for year_data in DATA]
+# Data
+years = ["2014", "2015", "2016", "2017", "2018", "2019", "2020", "2021", "2022", "2023"]
+buckets = ["1-1", "2-5", "6-10", "11-100", "101-1000", "1001-30000000"]
+data = {
+    "2014": [17.410227605477868, 16.11176217183986, 15.632757662414805, 15.446116676532212, 16.716943171826703, 18.156821563322765],
+    "2015": [17.446573602753478, 16.14852530113782, 15.627408549576069, 15.0055028132117, 15.565430373421485, 17.314701050452452],
+    "2016": [17.307221780905284, 16.297702171159543, 15.948641884223639, 14.799690714225637, 14.935989931859659, 16.09585768919658],
+    "2017": [17.338525603992114, 15.960924352297502, 15.912187993988933, 14.822102470001267, 14.778913482337416, 15.428145290012955],
+    "2018": [17.08551151136689, 16.187802102106698, 14.935072408852303, 14.832038213200583, 14.508674264491997, 14.800605964649103],
+    "2019": [16.818363305107052, 16.474269837858706, 14.944741674400241, 14.568394784374943, 14.690158822673334, 15.990949424635108],
+    "2020": [16.98821894111693, 15.936494557783181, 14.79960386342691, 14.435682562274105, 14.58651834886038, 15.869365567783806],
+    "2021": [17.125795647512877, 15.780419457145868, 14.631430892394002, 14.276477514399625, 14.337146941773641, 15.872474774329305],
+    "2022": [16.573462144306383, 15.283018703313582, 14.378277745163881, 14.0611924390084, 13.9886330091318, 15.769421394877273],
+    "2023": [15.4293630385597, 14.608379914730168, 14.118271697056592, 13.880215644749589, 13.767106666731275, 15.05749135510839]
+}
 
-# Create the figure
+# Create a line plot for each bucket
 fig = go.Figure()
 
-# Add a trace for each range (bucket)
-for i, range_label in enumerate(ranges):
-    values = [year_values[i] for year_values in all_values]
-    fig.add_trace(go.Scatter(x=years, y=values, mode='lines+markers', name=range_label))
+for i, bucket in enumerate(buckets):
+    bucket_values = [data[year][i] for year in years]
+    fig.add_trace(go.Scatter(x=years, y=bucket_values, mode='lines+markers', name=bucket))
 
 # Update layout
 fig.update_layout(
-    title="Perplexity over Time by Buckets",
+    title="Average Perplexity Over Years by Bucket",
     xaxis_title="Year",
-    yaxis_title="Perplexity",
+    yaxis_title="Average Perplexity",
     legend_title="Buckets",
     hovermode="x unified"
 )
 
-# Show the plot
-
-Perplexity_Across_Different_years_graph = fig
+# Show plot
+graph2 = fig
 
 #graph 3 tbd
 
@@ -794,7 +789,7 @@ perp1_div = Div(
             H3("Perplexity vs Years"),
             P("Taking the same data, we can convert it into a graph indicating the yearly trend. For most buckets, the average perplexity of dumps from more recent years seem to be lower than that of former years."),
             Img(src="images/prep-across-diff-year-global-dup-buckets.png", height = "300", width = "600" ),
-            plotly2fasthtml(Perplexity_Across_Different_years_graph),
+            plotly2fasthtml(graph2),
         ),
     Section(
             H3("Perplexity vs Document Duplication"),