dottyz

from datetime import timedelta
from fuzzywuzzy import fuzz

import matplotlib.pyplot as plt
import pandas as pd
import requests
import seaborn as sns

import json
import os

dottyz

# Identify the date structure used by each of the files as a dict:
#      * Key: data file name
#      * Value: [datetime format, hour difference bewteen timezone used and Eastern timezone]
date_formats = {
    'Bikeshare Ridership (2017 Q1).csv': ['%d/%m/%Y %H:%M', -4],
    'Bikeshare Ridership (2017 Q2).csv': ['%d/%m/%Y %H:%M', -4],
    'Bikeshare Ridership (2017 Q3).csv': ['%m/%d/%Y %H:%M', 0],
    'Bikeshare Ridership (2017 Q4).csv': ['%m/%d/%y %H:%M:%S', 0],
}
df = pd.DataFrame() # Initiate an empty DataFrame

dottyz

stations_start = df[['from_station_id', 'from_station_name']]
stations_end = df[['to_station_id', 'to_station_name']]
stations_start.columns = stations_end.columns = ['station_id', 'name']

# Extracts the unique station ID and name combination from the from_station and to_station columns
stations = pd.concat([stations_start, stations_end]).dropna(how='all').drop_duplicates().reset_index(drop=True)

dottyz

# Separate the stations without station IDs
no_ids = stations[stations['station_id'].isnull()]
for idx, miss in no_ids.iterrows():
    max_score = 0
    
    # Compare the similarity of the station without ID to each station in the API data
    for i, exist in bikeshare_stations[['station_id', 'name']].iterrows():
        score = fuzz.ratio(miss['name'], exist['name'])
        
        if score > 80 and score > max_score:

dottyz

stations = pd.concat([stations[~stations['station_id'].isnull()], no_ids])\
             .merge(bikeshare_stations[['station_id', 'lat', 'lon']], how='inner', on='station_id')\
             .drop_duplicates()
             
df = df.merge(stations, how='inner', left_on='from_station_name', right_on='name') \
       .merge(stations, how='inner', left_on='to_station_name', right_on='name', suffixes=['_from', '_to']) \
       .drop_duplicates()

df = df[[x for x in df.columns if not x.endswith('_station_id') and not x.endswith('_station_name') and x != 'trip_stop_time']]

dottyz

fig, ax = plt.subplots(1, 1, figsize=(16, 9))
sns.distplot(df['trip_duration_seconds'], hist=False, ax=ax)

dottyz

# Removing false start trips
df = df[(df['trip_duration_seconds']>=60)]

# Removing outliers
q1 = df['trip_duration_seconds'].quantile(0.25)
q3 = df['trip_duration_seconds'].quantile(0.75)

interquartile_range = q3 - q1

df = df[~((df['trip_duration_seconds'] < (q1 - 1.5 * interquartile_range)) \

dottyz

df.to_csv('./data/bikeshare_ridership.csv', index=False)

dottyz

import datetime as dt
import re

import numpy as np
import pandas as pd
import matplotlib.pyplot as plt
import seaborn as sns

from geopy import distance
from pandas.api.types import CategoricalDtype

dottyz

# Clean up column names for ease of use
df.columns = [' '.join(x.replace('trip_', '').replace('_seconds', '').split('_')).title() for x in df.columns]

df['Start Time'] = pd.to_datetime(df['Start Time'])

df['Date'] = df['Start Time'].apply(lambda x: x.strftime('%Y-%m-%d'))
df['Quarter'] = df['Start Time'].apply(lambda x: int((int(x.strftime('%m')) - 1) / 3) + 1)
df['Month'] = df['Start Time'].apply(lambda x: x.strftime('%B')).astype(month_type)
df['Day of Week'] = df['Start Time'].apply(lambda x: x.strftime('%a')).astype(day_type)
df['Hour'] = df['Start Time'].apply(lambda x: x.strftime('%H'))