Merge pull request #85 from opensafely/update-local-actions

Update local actions

.gitignore

@@ -4,6 +4,7 @@ model.log
 __pycache__
 .python-version
 /output/*
+released_outputs/*
 metadata/*
 venv/
 .DS_Store

analysis/report/panel_plots.py

@@ -378,7 +378,7 @@ def get_group_chart(
         rows = rows + 1
 
     # NOTE: constrained_layout=True available in matplotlib>=3.5
-    figure = plt.figure(figsize=(12 + 6 * (columns - 1), 4.8 * rows))
+    figure = plt.figure(figsize=(12 + 6 * (columns - 1), 4.8 * rows), dpi=300)
 
     lgds = []
     for index, panel in enumerate(groups):
@@ -395,6 +395,16 @@ def get_group_chart(
             repeated,
             autolabel=True,
         )
+
+        if title =="Sore Throat Tonsillitis":
+            title = "Sore Throat/Tonsillitis"
+        elif title == "Invasive Strep A":
+            title = "iGAS"
+        elif title == "Age Band":
+            title = "Age Band (years)"
+        elif title == "Imd":
+            title = "Index of Multiple Deprivation (IMD) quintile"
+
         ax.set_title(title)
         # Filter out group, but ignore case
         if exclude_group:
@@ -424,6 +434,7 @@ def get_group_chart(
                 lgd_params,
                 hide_legend,
             )
+
             # Save the season table only if there is more than one group
             more_than_one_group = panel_group_data.group.nunique() > 1
             if (produce_season_table and more_than_one_group) or mark_seasons:
@@ -432,7 +443,7 @@ def get_group_chart(
                     "group",
                     column_to_plot,
                     output_dir,
-                    panel_group_data.iloc[0]["name"],
+                    panel_group_data.iloc[0]["name"].replace("/", "_"),
                 )
                 if mark_seasons:
                     annotate_seasons(season_table, column_to_plot, ax)
@@ -677,7 +688,7 @@ def main():
         produce_season_table=produce_season_table,
         mark_seasons=mark_seasons,
     )
-    write_group_chart(chart, lgds, output_dir / output_name, plot_title)
+    write_group_chart(chart, lgds, output_dir / output_name)
     chart.close()
 
 

analysis/report/pcnt_table.py

@@ -2,6 +2,9 @@
 import argparse
 import pandas
 import fnmatch
+import re
+import matplotlib.dates as mdates
+import matplotlib.pyplot as plt
 
 from report_utils import ci_95_proportion, ci_to_str
 
@@ -38,6 +41,48 @@ def subset_table(measure_table, measures_pattern, date):
 
     return table_subset
 
+def plot_pcnt_over_time(measure_table, output_dir, type):
+    """
+    Plot the percentage over time for each measure.
+    """
+    measure_table['date'] = pandas.to_datetime(measure_table['date'])
+    measure_table['value'] *= 100
+
+
+    valid_dates = measure_table.dropna(subset=['value'])['date']
+    start_date = valid_dates.min()
+    end_date = valid_dates.max()
+    measure_table = measure_table[(measure_table['date'] >= start_date) & (measure_table['date'] <= end_date)]
+
+    plt.figure(figsize=(10, 6))
+    for name, group in measure_table.groupby('name'):
+
+        legend_label = name
+
+        if type == "clinical":
+            match = re.search(r"event_(\w+)_with_clinical_any_pcnt", name)
+            if match:
+                legend_label = match.group(1).title()
+        elif type == "medication":
+            match = re.search(r"event_(\w+)_with_medication_any_pcnt", name)
+            if match:
+                legend_label = match.group(1).title()
+
+        plt.plot(group['date'], group['value'], label=legend_label)
+
+    plt.xlabel('Date')
+    plt.ylabel('Percentage')
+    plt.title(f'Percentage with {type} event')
+    plt.legend(loc='center left', bbox_to_anchor=(1, 0.5), fontsize='small')
+    plt.xticks(rotation=90)
+    plt.grid(True)
+    plt.ylim(bottom=0)
+    plt.xlim(start_date, end_date)
+    plt.gca().xaxis.set_major_locator(mdates.MonthLocator())
+    plt.gca().xaxis.set_major_formatter(mdates.DateFormatter('%b-%Y'))
+    plt.tight_layout()
+    plt.savefig(output_dir / f"pcnt_over_time_{type}.jpg", dpi=300)
+    plt.close()
 
 def match_paths(files, pattern):
     return fnmatch.filter(files, pattern)
@@ -106,6 +151,17 @@ def main():
     table = pandas.DataFrame(joined)
     table.to_html(output_dir / "pcnt_with_indication.html", index=True)
 
+    filtered_table_med = measure_table[
+        measure_table['name'].str.contains('with_medication_any_pcnt')
+    ]   
+    plot_pcnt_over_time(filtered_table_med, output_dir, "medication")
+
+    filtered_table_clinical = measure_table[
+        measure_table['name'].str.contains('with_clinical_any_pcnt')
+    ]    
+    plot_pcnt_over_time(filtered_table_clinical, output_dir, "clinical")
+
+
 
 if __name__ == "__main__":
     main()

analysis/report/report.ipynb

@@ -119,12 +119,12 @@
     "display(Markdown(\"The below charts show the count of patients prescribed any of the above listed antibiotics each month, followed by a table with the underlying counts and a chart with of the rate of prescribing with the years stacked on top of each other.\"))\n",
     "display_image(\"medications_bar_measures_count.jpeg\")\n",
     "display_table(\"event_counts_medication.csv\")\n",
-    "display_image(\"all_medications_by_year.png\")\n",
+    "display_image(\"all_medications_by_year.jpg\")\n",
     "display(Markdown(\"The below charts show the monthly count and rate of patients with any of the listed antibiotics across the study period, with a breakdown by key demographic subgroups.\"))\n",
     "display(Markdown(\"##### Count\"))\n",
-    "display_image(\"medication_any_by_subgroup_count.png\")\n",
+    "display_image(\"medication_any_by_subgroup_count.jpg\")\n",
     "display(Markdown(\"##### Rate\"))\n",
-    "display_image(\"medication_any_by_subgroup.png\")"
+    "display_image(\"medication_any_by_subgroup.jpg\")"
    ]
   },
   {
@@ -294,12 +294,12 @@
     "display(Markdown(\"The below charts show the count of patients with any of the above listed clinical events each month, followed by a table with the underlying counts and a chart with the rate of clinical events with the years stacked on top of each other.\"))\n",
     "display_image(\"clinical_bar_measures_count.jpeg\")\n",
     "display_table(\"event_counts_clinical.csv\")\n",
-    "display_image(\"any_clinical_event_by_year.png\")\n",
+    "display_image(\"any_clinical_event_by_year.jpg\")\n",
     "display(Markdown(\"The below charts show the monthly count and rate of patients with any of the listed clinical events across the study period, with a breakdown by key demographic subgroups.\"))\n",
     "display(Markdown(\"##### Count\"))\n",
-    "display_image(\"clinical_any_by_subgroup_count.png\")\n",
+    "display_image(\"clinical_any_by_subgroup_count.jpg\")\n",
     "display(Markdown(\"##### Rate\"))\n",
-    "display_image(\"clinical_any_by_subgroup.png\")"
+    "display_image(\"clinical_any_by_subgroup.jpg\")"
    ]
   },
   {

analysis/report/report_utils.py

@@ -428,12 +428,12 @@ def subset_table(measure_table, measures_pattern, measures_list):
     return measure_table[measure_table["name"].isin(measures_list)]
 
 
-def write_group_chart(group_chart, lgds, path, plot_title):
-    suptitle = plt.suptitle(plot_title)
-    group_chart.savefig(
-        path, bbox_extra_artists=tuple(lgds) + (suptitle,), bbox_inches="tight"
-    )
-
+def write_group_chart(group_chart, lgds, path, plot_title=None):
+    save_args = {"fname": f"{path}.jpg", "dpi": 300, "bbox_inches": "tight"}
+    if plot_title:
+        suptitle = plt.suptitle(plot_title)
+        save_args.update({"bbox_extra_artists": (tuple(lgds) + (suptitle,))})
+    group_chart.savefig(**save_args)
 
 # NOTE: These paths will only work for notebook generation, which is run on /workspace
 REPORT_DIR = Path.cwd().parent.parent / "output/report"
@@ -551,7 +551,7 @@ def display_medicine(
         )
     )
     display_image(
-        f"event_code_{medicine_path}_rate_top_5_codes_over_time.png",
+        f"event_code_{medicine_path}_rate_top_5_codes_over_time.jpg",
         dir=results_dir,
     )
     display(
@@ -560,9 +560,9 @@ def display_medicine(
         )
     )
     display(Markdown("##### Count"))
-    display_image(f"{medicine_path}_by_subgroup_count.png", dir=results_dir)
+    display_image(f"{medicine_path}_by_subgroup_count.jpg", dir=results_dir)
     display(Markdown("##### Rate"))
-    display_image(f"{medicine_path}_by_subgroup.png", dir=results_dir)
+    display_image(f"{medicine_path}_by_subgroup.jpg", dir=results_dir)
     if time_period == "month":
         display(
             Markdown(
@@ -575,7 +575,7 @@ def display_medicine(
             )
         )
         display_image(
-            f"{medicine_path}_with_clinical_any_by_subgroup.png",
+            f"{medicine_path}_with_clinical_any_by_subgroup.jpg",
             dir=results_dir,
         )
 
@@ -625,7 +625,7 @@ def display_clinical(
             )
         )
         display_image(
-            f"event_code_{clinical_path}_rate_top_5_codes_over_time.png",
+            f"event_code_{clinical_path}_rate_top_5_codes_over_time.jpg",
             dir=results_dir,
         )
         display(
@@ -635,10 +635,10 @@ def display_clinical(
         )
         display(Markdown("##### Count"))
         display_image(
-            f"{clinical_path}_by_subgroup_count.png", dir=results_dir
+            f"{clinical_path}_by_subgroup_count.jpg", dir=results_dir
         )
         display(Markdown("##### Rate"))
-        display_image(f"{clinical_path}_by_subgroup.png", dir=results_dir)
+        display_image(f"{clinical_path}_by_subgroup.jpg", dir=results_dir)
         if time_period == "month":
             display(Markdown("##### Rate with an antibiotic of interest"))
             display(
@@ -647,6 +647,6 @@ def display_clinical(
                 )
             )
             display_image(
-                f"{clinical_path}_with_medication_any_by_subgroup.png",
+                f"{clinical_path}_with_medication_any_by_subgroup.jpg",
                 dir=results_dir,
             )

analysis/report/report_weekly.ipynb

@@ -124,9 +124,9 @@
     "display_table(\"event_counts_medication.csv\", dir=WEEKLY_RESULTS_DIR)\n",
     "display(Markdown(\"The below charts show the weekly count and rate of patients with any of the listed antibiotics across the study period, with a breakdown by key demographic subgroups.\"))\n",
     "display(Markdown(\"##### Count\"))\n",
-    "display_image(\"medication_any_by_subgroup_count.png\", dir=WEEKLY_RESULTS_DIR)\n",
+    "display_image(\"medication_any_by_subgroup_count.jpg\", dir=WEEKLY_RESULTS_DIR)\n",
     "display(Markdown(\"##### Rate\"))\n",
-    "display_image(\"medication_any_by_subgroup.png\", dir=WEEKLY_RESULTS_DIR)"
+    "display_image(\"medication_any_by_subgroup.jpg\", dir=WEEKLY_RESULTS_DIR)"
    ]
   },
   {
@@ -283,9 +283,9 @@
     "display_table(\"event_counts_clinical.csv\", dir=WEEKLY_RESULTS_DIR)\n",
     "display(Markdown(\"The below charts show the weekly count and rate of patients with any of the listed clinical events across the study period, with a breakdown by key demographic subgroups.\"))\n",
     "display(Markdown(\"##### Count\"))\n",
-    "display_image(\"clinical_any_by_subgroup_count.png\", dir=WEEKLY_RESULTS_DIR)\n",
+    "display_image(\"clinical_any_by_subgroup_count.jpg\", dir=WEEKLY_RESULTS_DIR)\n",
     "display(Markdown(\"##### Rate\"))\n",
-    "display_image(\"clinical_any_by_subgroup.png\", dir=WEEKLY_RESULTS_DIR)"
+    "display_image(\"clinical_any_by_subgroup.jpg\", dir=WEEKLY_RESULTS_DIR)"
    ]
   },
   {

analysis/report/top_5_over_time.py

@@ -0,0 +1,42 @@
+import argparse
+import pandas as pd
+from pathlib import Path
+
+from top_5_report import plot_top_codes_over_time
+
+
+def parse_args():
+    parser = argparse.ArgumentParser()
+    parser.add_argument(
+        "--input-file",
+        required=True,
+        help="Path to single joined measures file",
+    )
+    parser.add_argument(
+        "--output-dir",
+        required=True,
+        type=Path,
+        help="Path to the output directory",
+    )
+    parser.add_argument(
+        "--frequency",
+        default="month",
+        choices=["month", "week"],
+        help="The frequency of the data",
+    )
+    parser.add_argument(
+        "--xtick-frequency",
+        help="Display every nth xtick",
+        type=int,
+        default=1,
+    )
+    return parser.parse_args()
+
+
+def main():
+    args = parse_args()
+    df = pd.read_csv(args.input_file)
+    plot_top_codes_over_time(df, "sore_throat_tonsillitis_rate", args.output_dir, args.frequency, args.xtick_frequency)
+
+if __name__ == "__main__":
+    main()

analysis/report/top_5_report.py

@@ -324,8 +324,9 @@ def plot_top_codes_over_time(
     plt.legend(bbox_to_anchor=(1.05, 1), loc="upper left")
 
     plt.savefig(
-        output_dir / f"{measure}_top_5_codes_over_time.png",
+        output_dir / f"{measure}_top_5_codes_over_time.jpg",
         bbox_inches="tight",
+        dpi=300,
     )