analysis for the movie graph

scripts/movie_bench/closeness_results/analysis.py

@@ -0,0 +1,28 @@
+import os
+import matplotlib.pyplot as plt
+import numpy as np
+import pandas as pd
+
+dfs = {
+    i: pd.read_csv(f"top_movies_{i:02d}_c.txt", sep='\t', usecols=[1], names=["actor"])
+    for i in [5] + list(range(10, 71, 10))}
+sets = {i: set(df["movie"]) for i, df in dfs.items()}
+
+diff = []
+for i in sets.keys():
+    diff.append([len(sets[i]) - len(sets[i] & sets[j]) for j in sets.keys()])
+diff = np.array(diff, dtype=float)
+diff /= len(next(iter(sets.values())))
+
+plt.matshow(diff)
+for (i, j), z in np.ndenumerate(diff):
+    plt.gca().text(j, i, f'{z:0.2f}', ha='center', va='center')
+plt.gca().set_xticks(np.linspace(0.0, len(sets) - 1, len(sets)))
+plt.gca().set_yticks(np.linspace(0.0, len(sets) - 1, len(sets)))
+plt.gca().set_xticklabels([f"{i:d}" for i in sets.keys()])
+plt.gca().set_yticklabels([f"{i:d}" for i in sets.keys()])
+plt.ylabel("\nNumber of Votes")
+plt.xlabel("\nNumber of Votes")
+cb = plt.colorbar()
+cb.set_label("\npercentace of difference in results varing the number of votes")
+plt.show()