|
| 1 | +import pandas as pd |
| 2 | + |
| 3 | +from api_data import ApiDataInterface |
| 4 | +from web_data import ColumnNames, WebDataGenerator, RowDataGetter |
| 5 | + |
| 6 | +WEB_BASE_DIR = 'web' |
| 7 | +DATA_BASE_DIR = 'data/maryland' |
| 8 | + |
| 9 | +COLUMNS = ['REPORT_NO', 'REPORT_TYPE', 'HARM_EVENT_DESC1', 'HARM_EVENT_DESC2', 'LATITUDE', 'LONGITUDE', 'YEAR'] |
| 10 | + |
| 11 | +HARM = { |
| 12 | + 'Pedestrian': 'ped', |
| 13 | + 'Bicycle': 'bike', |
| 14 | +} |
| 15 | + |
| 16 | +COLUMN_NAMES = ColumnNames(latitude='LATITUDE', longitude='LONGITUDE', year='YEAR') |
| 17 | + |
| 18 | + |
| 19 | +class MarylandApiDataInterface(ApiDataInterface): |
| 20 | + def convert_to_df(self): |
| 21 | + df = pd.read_csv('data/Maryland_Statewide_Vehicle_Crashes.csv') |
| 22 | + df.to_pickle('data/maryland.pkl') |
| 23 | + |
| 24 | + def filter_data(self): |
| 25 | + df = pd.read_pickle(f'data/maryland.pkl') |
| 26 | + for col in df.columns: |
| 27 | + df[col] = pd.to_numeric(df[col], errors='ignore') |
| 28 | + df = df[COLUMNS] |
| 29 | + df = df.loc[(df['REPORT_TYPE'] == 'Fatal Crash') | (df['REPORT_TYPE'] == 'Injury Crash')] |
| 30 | + df.to_pickle(f'data/maryland-filtered.pkl') |
| 31 | + |
| 32 | + def read_data(self): |
| 33 | + return pd.read_pickle(f'data/maryland-filtered.pkl') |
| 34 | + |
| 35 | + |
| 36 | +class MarylandRowDataGetter(RowDataGetter): |
| 37 | + @staticmethod |
| 38 | + def item_id(row): |
| 39 | + return row['REPORT_NO'] |
| 40 | + |
| 41 | + @staticmethod |
| 42 | + def category(row): |
| 43 | + return HARM.get(row['HARM_EVENT_DESC1'], 'car') |
| 44 | + |
| 45 | + |
| 46 | +def generate_web_data(): |
| 47 | + df = MarylandApiDataInterface().read_data() |
| 48 | + |
| 49 | + web_data_generator = WebDataGenerator(row_data_getter=MarylandRowDataGetter(), |
| 50 | + column_names=COLUMN_NAMES, |
| 51 | + state='maryland') |
| 52 | + web_data_generator.iterate_and_save(df, latlong_interval=2) |
| 53 | + |
| 54 | + |
| 55 | +if __name__ == '__main__': |
| 56 | + # convert_to_df() |
| 57 | + # filter_data() |
| 58 | + # df = pd.read_pickle(f'data/maryland-filtered.pkl') |
| 59 | + print('Generating web data...') |
| 60 | + generate_web_data() |
| 61 | + print('Done.') |
0 commit comments