Complete Assignment 3

Assignment3/.gitignore

@@ -1 +1,6 @@
 data/*.csv
+
+!data/r1.csv
+!data/r2.csv
+!data/r3.csv
+!data/r4.csv

Assignment3/code/plot.py

@@ -0,0 +1,57 @@
+import pandas as pd
+import matplotlib.pyplot as plt
+import os
+
+CHOSEN_DATE: str = '2018-09-01'
+plt.rcParams['font.sans-serif'] = ['SimHei']
+
+def get_path(filename: str) -> str:
+    return os.path.join(os.path.dirname(os.getcwd()), 'data', filename)
+
+
+def read_csv() -> pd.DataFrame:
+    r5 = pd.read_csv(get_path("r5.csv"))
+    r5['start_time'] = pd.to_datetime(r5['start_time'])
+    r5['end_time'] = pd.to_datetime(r5['end_time'])
+
+    return r5
+
+
+def choose_section(r5: pd.DataFrame) -> tuple[str, str, str]:
+    sections = r5.groupby('section')['percentage'].mean().sort_values(ascending=False)
+    selected_sections = (sections.index[0], sections.index[len(sections) // 3], sections.index[len(sections) // 2])
+    return selected_sections
+
+
+def plot(r5: pd.DataFrame, selected_sections: tuple[str, str, str]) -> None:
+    plt.figure(figsize=(15, 8))
+
+    colors = ('red', 'green', 'blue')
+    for i, section in enumerate(selected_sections):
+        section_data: pd.Series = r5[r5['section'] == section]
+        daily_data: pd.Series = section_data[section_data['start_time'].dt.strftime('%Y-%m-%d') == CHOSEN_DATE]
+        final_data: pd.Series = daily_data[['start_time', 'end_time', 'percentage']]
+
+        plt.plot(final_data['start_time'],
+                 final_data['percentage'],
+                 marker='o',
+                 color=colors[i],
+                 label=section)
+
+    plt.title(f'Three Sections Percentage in {CHOSEN_DATE}')
+    plt.xlabel('Time')
+    plt.ylabel('Percentage(%)')
+    plt.grid(True)
+    plt.legend()
+    plt.xticks(rotation=45)
+
+    plt.tight_layout()
+
+    plt.savefig(get_path('parking_utilization.png'))
+    plt.show()
+
+
+if __name__ == '__main__':
+    result5 = read_csv()
+    selected = choose_section(result5)
+    plot(result5, selected)

Assignment3/code/requirements.txt

@@ -0,0 +1,7 @@
+# main packages
+pyspark~=3.5.3
+ipykernel~=6.29.5
+
+# for task 5 subtask only
+matplotlib~=3.9
+pandas~=2.2.3

Assignment3/data/r1.csv

@@ -0,0 +1,44 @@
+section,count
+科技南一路,92
+高新南九道,26
+创业路(南油段),72
+文心四路,24
+高新南七道,41
+招商路(蛇口西段),61
+科技南十路,29
+后海大道辅道(蛇口北段),58
+登良路,29
+学府路(前海段),11
+工业九路,50
+文心三路,12
+文心一路,13
+荔园路(蛇口西段),70
+荔园路(蛇口东段),54
+高新南环路,240
+四海路,55
+科技南路,87
+文心五路,14
+海德二道,42
+中心路(后海湾段),63
+金世纪路,19
+创业路辅道,43
+文心六路,58
+科技南十二路,10
+高新南十一道,60
+高新南十道,73
+龙城路,31
+南头街,21
+海月路(东段),19
+高新南五道,18
+海德一道,51
+高新南三道,22
+南新路,23
+海德二道（南油段）,45
+文心二路,30
+海德三道,46
+常兴路,66
+创业路(后海湾段),18
+华明路,3
+高新南四道,26
+后海大道辅道(后海段),80
+海德一道(后海湾段),26