Merge pull request #1111 from cal-itp/holiday-service-update

Holiday service update
cal-itp · May 13, 2024 · b1e5d4f · b1e5d4f
2 parents 1ce0303 + 5b8e2b7
commit b1e5d4f
Show file tree

Hide file tree

Showing 4 changed files with 1,201 additions and 417 deletions.
diff --git a/holiday_service_research/funcs_vars.py b/holiday_service_research/funcs_vars.py
@@ -1,4 +1,7 @@
 import matplotlib.pyplot as plt
+import matplotlib.colors as mcolors
+from matplotlib.colors import ListedColormap
+
 import numpy as np
 import pandas as pd
 import seaborn as sns  # For visualization (optional)
@@ -56,10 +59,10 @@
         'date': '2024-01-15',
     }, {
         'name': "Reference Weekday",
-        'date': '2023-12-14',
+        'date': '2023-12-15',
     }, {
         'name': "Reference Saturday",
-        'date': '2023-11-25',
+        'date': '2023-12-16',
     }, {
         'name': "Reference Sunday",
         'date': '2023-12-17',
@@ -88,21 +91,93 @@
 
 def plot_confusion_matrices(df, y_true, y_pred, title): 
     desired_order = ['No service', 'Reduced service', 'Regular service']
+    x_desired_order = ['No service', 'Reduced service', 'Regular service']
+    y_desired_order = [ 'Regular service', 'Reduced service', 'No service']
     cm = confusion_matrix(y_true=df[y_true], y_pred=df[y_pred], labels=desired_order)
     df_cm = pd.DataFrame(cm, index=desired_order, columns=desired_order)
-    # df_cm = df_cm.reindex(desired_order, axis=0)  # Rows
-    # df_cm = df_cm.reindex(desired_order, axis=1)  # Columns
+    df_cm = df_cm.reindex(y_desired_order, axis=0)  # Rows
+    df_cm = df_cm.reindex(x_desired_order, axis=1)  # Columns
+    df_cm = (df_cm/df_cm.sum().sum()).round(2)*100 # Make cm based on percentages
+
     # https://stackoverflow.com/questions/64800003/seaborn-confusion-matrix-heatmap-2-color-schemes-correct-diagonal-vs-wrong-re
     vmin = np.min(cm)
     vmax = np.max(cm)
-    off_diag_mask = np.eye(*cm.shape, dtype=bool)
+    #It might have been easier to make this manually :P. Make a diagonal matrix from upper left to lower right, then flip it on tahe horizontal.
+    gtfs_matches_website = np.fliplr(np.eye(*cm.shape, dtype=bool, k=0))
+    gtfs_greater_website = ([
+        [True, True,  False],
+        [True, False, False],
+        [False, False, False]])
+    gtfs_less_website = ([
+        [False, False,  False],
+        [False,  False, True],
+        [False, True, True]])
+
+    gtfs_greater_website = pd.DataFrame(gtfs_greater_website).to_numpy()
+    gtfs_less_website = pd.DataFrame(gtfs_less_website).to_numpy()
+    plt.rcParams.update({'font.size': 13})
+
+    # Used https://redketchup.io/color-picker to match colors in comparison graph to make this heatmap
+    color = mcolors.to_rgb('#1F77B4')
+    blue_cmap = mcolors.ListedColormap([color])
+    color = mcolors.to_rgb('#FF7F0E')
+    orange_cmap = mcolors.ListedColormap([color])
+    color = mcolors.to_rgb('#2CA02C')
+    light_green_cmap = mcolors.ListedColormap([color])
 
     plt.figure(figsize=(8, 6))
-    sns.heatmap(df_cm, annot=True,  fmt='g', mask=~off_diag_mask, cmap="Blues", vmin=0, vmax=1, cbar=False, linewidths=0.8, linecolor='k')
-    sns.heatmap(df_cm, annot=True,  fmt='g', mask=off_diag_mask, cmap="OrRd", vmin=0, vmax=1, cbar=False, linewidths=0.8, linecolor='k')
-    plt.xlabel('Service Level on Website')
-    plt.ylabel('GTFS Service Levels')
-    plt.title(title)
-    # plt.show()
+    font_size = {"fontsize":18}
+    sns.heatmap(df_cm, annot=True,  annot_kws=font_size, fmt='g', mask=~gtfs_matches_website, cmap=blue_cmap, vmin=0, vmax=1, cbar=False, linewidths=0.8, linecolor='k')
+    sns.heatmap(df_cm, annot=True,  annot_kws=font_size, fmt='g', mask=~gtfs_greater_website, cmap=light_green_cmap, vmin=.5, vmax=.6, cbar=False, linewidths=0.8, linecolor='k')
+    sns.heatmap(df_cm, annot=True,  annot_kws=font_size, fmt='g', mask=~gtfs_less_website, cmap=orange_cmap, vmin=.04, vmax=.06, cbar=False, linewidths=0.8, linecolor='k')
+
+    #fontsize=14.0
+    plt.xlabel('Service Level on Website (% of agencies)', fontweight='bold')
+    plt.ylabel('GTFS Service Levels (% of agencies)', fontweight='bold')
+    plt.title(title, fontweight='bold')
     file = title
-    plt.savefig(f"plots/{file}.png")
+    plt.savefig(f"plots/{file}.png")
+    # return cm, df_cm
+
+excel_col_order = ['Name', 'Notes', 'gtfs_dataset_name',
+'Total VOMS (NTD) (from Provider)', 'Customer Facing',"name",
+"Reference Saturday",
+"Reference Sunday",
+"Reference Weekday",
+"Holiday Schedule – Thanksgiving Day",
+"Thanksgiving Day",
+"score - Thanksgiving Day",
+"score_text - Thanksgiving Day",
+"Holiday Schedule – Christmas Day",
+"Christmas Day",
+"score - Christmas Day",
+"score_text - Christmas Day",
+"Holiday Schedule – New Year's Day",
+"New Year's Day",
+"score - New Year's Day",
+"score_text - New Year's Day",
+"Holiday Schedule – MLK Day",
+"MLK Day",
+"score - MLK Day",
+"score_text - MLK Day",
+"Holiday Schedule – Veterans Day (Observed)",
+"Veterans Day (Observed)",
+"score - Veterans Day (Observed)",
+"score_text - Veterans Day (Observed)",
+"Holiday Schedule – Veterans Day",
+"Veterans Day",
+"score - Veterans Day",
+"score_text - Veterans Day",
+"Holiday Schedule – Day after Thanksgiving Day",
+"Day After Thanksgiving",
+"score - Day After Thanksgiving",
+"score_text - Day After Thanksgiving",
+"Holiday Schedule – Christmas Eve",
+"Christmas Eve",
+"score - Christmas Eve",
+"score_text - Christmas Eve",
+"Holiday Schedule – New Year’s Eve",
+"New Year's Eve",
+"score - New Year's Eve",
+"score_text - New Year's Eve",
+"Holiday Schedule Notes"]