💄 Adds ticks to int grades

analyze.py

@@ -27,27 +27,32 @@ def analyze(name: str, data: List[np.ndarray]):
     return F, p
 
 
-def plot_box(data, labels, Fs, ps, title, titles):
+def plot_box(data, labels, Fs, ps, title):
     if not graph:
         return
 
+    titles = ["Priemer", "Matematika", "Slovenčina", "Angličtina"]
+
     fig, axs = plt.subplots(2, 2, sharex=True)
     fig.suptitle(title)
     fig.set_size_inches(12, 9)
 
-    for i in range(2):
     for j in range(2):
-            print(f"{i}x{j} giving {i * 2 + j}")
-            axs[i, j].boxplot(data[i * 2 + j], labels=labels)
-            axs[i, j].set_title(titles[i * 2 + j])
+        for k in range(2):
+            index = j * 2 + k
+            axs[j, k].boxplot(data[index], labels=labels)
+            axs[j, k].set_title(titles[index])
 
-            F = round(Fs[i * 2 + j], 2)
-            p = round(ps[i * 2 + j], 4)
-            axs[i, j].text(0.01, 0.99, f"F-stat: {F}\np-val: {p}", ha="left", va="top", transform=axs[i, j].transAxes,
+            if index > 0:
+                axs[j, k].set_yticks(np.arange(1, 6, 1))
+
+            F = round(Fs[index], 2)
+            p = round(ps[index], 4)
+            axs[j, k].text(0.01, 0.99, f"F-stat: {F}\np-val: {p}", ha="left", va="top", transform=axs[j, k].transAxes,
                            fontweight="bold")
 
-            avgs = np.array([a.mean() for a in data[i * 2 + j]])
-            print(avgs)
+            medians = np.array([np.median(a) for a in data[j * 2 + k]])
+            print(medians)  # TODO: add to graph
 
     fig.tight_layout()
     fig.show()

analyze_ses.py

@@ -1,6 +1,6 @@
 import numpy as np
 
-from analyze import analyze
+from analyze import analyze, plot_box
 
 dataset = np.load("clean.npy")
 print(f"dataset shape: {dataset.shape}; analyzing column 6 (ses)")
@@ -12,14 +12,19 @@ print("")
 
 def analyze_ses(name: str, col: np.ndarray):
     sex_col = dataset[:, 6]
-    analyze(name, [
+    data = [
         col[sex_col == 0],
         col[sex_col == 1],
         col[sex_col == 2]
-    ])
+    ]
+    F, p = analyze(name, data)
+    return data, F, p
 
 
-analyze_ses("gpa", dataset[:, 2])
-analyze_ses("math", dataset[:, 3])
-analyze_ses("slovak", dataset[:, 4])
-analyze_ses("english", dataset[:, 5])
+data_gpa, F_gpa, p_gpa = analyze_ses("gpa", dataset[:, 2])
+data_math, F_math, p_math = analyze_ses("math", dataset[:, 3])
+data_slovak, F_slovak, p_slovak = analyze_ses("slovak", dataset[:, 4])
+data_english, F_english, p_english = analyze_ses("english", dataset[:, 5])
+
+plot_box([data_gpa, data_math, data_slovak, data_english], ["Nižšia trieda", "Stredná trieda", "Vyššia trieda"],
+         [F_gpa, F_math, F_slovak, F_english], [p_gpa, p_math, p_slovak, p_english], "Socio-ekonomická trieda")

analyze_sex.py

@@ -24,6 +24,5 @@ data_math, F_math, p_math = analyze_sex("math", dataset[:, 3])
 data_slovak, F_slovak, p_slovak = analyze_sex("slovak", dataset[:, 4])
 data_english, F_english, p_english = analyze_sex("english", dataset[:, 5])
 
-plot_box([data_gpa, data_math, data_slovak, data_english], ["Female", "Male"],
-         [F_gpa, F_math, F_slovak, F_english], [p_gpa, p_math, p_slovak, p_english],
-         "Pohlavie", ["Priemer", "Matematika", "Slovenčina", "Angličtina"])
+plot_box([data_gpa, data_math, data_slovak, data_english], ["Ženy", "Muži"],
+         [F_gpa, F_math, F_slovak, F_english], [p_gpa, p_math, p_slovak, p_english], "Pohlavie")