Functions used in series of analyses on the Austin Animal Center's animal intakes and outcomes datasets.

austin_animal_center_dataset_analysis_functions.py

import requests
import numpy as np
import pandas as pd
from six.moves.urllib.error import HTTPError


def get_soda_api_data(endpoint, count=1000, offset=0, return_df=True):
    params = {'$limit': count, '$offset': offset}
    
    results = []

    while True:

        try:
            r = requests.get(endpoint, params=params)
            rcontent = r.json()

            if rcontent == []:
                break

            results.append(rcontent)
            offset += count
            params['$offset'] = offset

        except HTTPError as err:

            if err.response.status_code == '404':
                break
            else:
                print(err.response.status_code)
    
    if return_df:
        results_df = pd.DataFrame()

        for i in results:
            results_df = results_df.append(pd.io.json.json_normalize(i))
        
        return results_df
    
    else:
        return results


def create_unique_id(df, event):
    df.reset_index(inplace=True)

    df[event + '_number'] = df.groupby(['animal_id'])[event + '_datetime'].rank(method='dense', ascending=False)
    df['animal_id_new'] = df['animal_id'] + '_' + df[event + '_number'].astype(int).astype(str)

    return df
    

def transform_age(df, age_column):
    df = df.loc[df.loc[:, age_column] != 'NULL']
    
    range_column = age_column + '_Period Range'
    
    df[age_column + '_Periods'], df[range_column] = df.loc[:, age_column].str.split(' ').str[0].fillna(0).astype(int), df.loc[:, age_column].str.split(' ').str[1].fillna(0)
    
    
    df[range_column] = np.where(df[range_column].str.contains('day'), 1, 
                                  np.where(df[range_column].str.contains('week'), 7, 
                                           np.where(df[range_column].str.contains('month'), 30, 
                                                    np.where(df[range_column].str.contains('year'), 365, 0)))).astype(int)

    df[age_column + '_(days)'] = df[range_column] * df[age_column + '_Periods']
    df[age_column + '_(years)'] = df[age_column + '_(days)'] / 365
    
    df[age_column + '_age_group'] = pd.cut(df[age_column + '_(years)'], 10)

    return df


def transform_date(df, event):
    event_date = event + '_datetime'
    df[event_date] = pd.to_datetime(df['datetime'])

    df[event + '_month'] = df[event_date].dt.month
    df[event + '_year'] = df[event_date].dt.year
    df[event + '_monthyear'] = df[event + '_datetime'].dt.to_period('M')
    
    df[event + '_weekday'] = df[event_date].dt.weekday_name
    df[event + '_hour'] = df[event_date].dt.hour
    
    return df


def transform_birthdates(df):
    df['date_of_birth'] = pd.to_datetime(df['date_of_birth'])

    df['dob_year'] = df['date_of_birth'].dt.year
    df['dob_month'] = df['date_of_birth'].dt.month
    df['dob_monthyear'] = pd.to_datetime(df['monthyear']).dt.to_period('M')

    return df