From 0795ab3439fb8a3c405b0971f1b48804c23b2981 Mon Sep 17 00:00:00 2001 From: natam1 Date: Fri, 31 May 2024 18:51:42 +0000 Subject: [PATCH] continue moving fx to utils --- sb125_analyses/corridor_study/_utils.py | 22 + .../data_downloads_skirball.ipynb | 1426 ++++++++--------- 2 files changed, 662 insertions(+), 786 deletions(-) diff --git a/sb125_analyses/corridor_study/_utils.py b/sb125_analyses/corridor_study/_utils.py index 5d84d4ba4..d93da68a2 100644 --- a/sb125_analyses/corridor_study/_utils.py +++ b/sb125_analyses/corridor_study/_utils.py @@ -80,6 +80,28 @@ def get_unique_agencies(agency_list): return df +def get_agencies_occurances(df): + + ##get list of columns that have agency names + columns_to_keep = list(df.columns.values) + + ## remove columns that are not agency names + unwanted_cols = {'transit_agency', 'unique_agencies', 'primary_mode', 'transit_submode', 'n', 'agency_count', 'n_modes_taken' } + columns_to_keep = [e for e in columns_to_keep if e not in unwanted_cols] + + df_agencies = df.loc[:, columns_to_keep] + + ## transpose the df again + df_agencies = pd.DataFrame(df_agencies).transpose().reset_index().rename(columns={'index':'agency'}) + + ##sum up the number of trips by agency + df_agencies['n_trips'] = df_agencies[list(df_agencies.columns)].sum(axis=1) + + ## filter df to only have two cols + df_agencies = df_agencies>>select(_.agency, _.n_trips) + + return df_agencies + """ Streetlight Analysis Utils """ diff --git a/sb125_analyses/corridor_study/data_downloads_skirball.ipynb b/sb125_analyses/corridor_study/data_downloads_skirball.ipynb index 91b7ce9b1..eff178baf 100644 --- a/sb125_analyses/corridor_study/data_downloads_skirball.ipynb +++ b/sb125_analyses/corridor_study/data_downloads_skirball.ipynb @@ -19,7 +19,7 @@ "name": "stderr", "output_type": "stream", "text": [ - "/tmp/ipykernel_515/2202862553.py:12: DeprecationWarning: Importing display from IPython.core.display is deprecated since IPython 7.14, please import from IPython display\n", + "/tmp/ipykernel_469/2202862553.py:12: DeprecationWarning: Importing display from IPython.core.display is deprecated since IPython 7.14, please import from IPython display\n", " from IPython.core.display import display\n" ] } @@ -108,7 +108,7 @@ "name": "stderr", "output_type": "stream", "text": [ - "/tmp/ipykernel_515/2643890524.py:1: DtypeWarning: Columns (18,19,20,25,26,28,29,30,31,33,36,38,39,40,41,42,43) have mixed types. Specify dtype option on import or set low_memory=False.\n", + "/tmp/ipykernel_469/2643890524.py:1: DtypeWarning: Columns (18,19,20,25,26,28,29,30,31,33,36,38,39,40,41,42,43) have mixed types. Specify dtype option on import or set low_memory=False.\n", " r_trips = to_snakecase(pd.read_csv(f\"{GCS_PATH}{replica_trips}\"))\n" ] } @@ -223,23 +223,23 @@ " \n", " \n", " \n", - " 423350\n", - " 5926084662673022461\n", - " 1 (Tract 2643.05, Los Angeles, CA)\n", - " 2643.05 (Los Angeles, CA)\n", + " 314707\n", + " 8505387565750628825\n", + " 1 (Tract 1349.07, Los Angeles, CA)\n", + " 1349.07 (Los Angeles, CA)\n", " Los Angeles County, CA\n", " California\n", - " 3 (Tract 9105.01, Los Angeles, CA)\n", - " 9105.01 (Los Angeles, CA)\n", + " 4 (Tract 6037.02, Los Angeles, CA)\n", + " 6037.02 (Los Angeles, CA)\n", " Los Angeles County, CA\n", " California\n", " private_auto\n", - " work\n", + " social\n", " home\n", - " 14:57:00\n", - " 16:25:56\n", - " 88\n", - " 57.2\n", + " 13:31:00\n", + " 14:24:59\n", + " 53\n", + " 31.2\n", " unknown_vehicle_type\n", " other_non_bev\n", " NaN\n", @@ -247,253 +247,253 @@ " NaN\n", " mixed_use\n", " multi_family\n", - " civic_institutional\n", - " civic_institutional\n", - " 9280162199592818201\n", - " 3819422246701548245\n", - " 59.0\n", + " single_family\n", + " single_family\n", + " 13590816709645713585\n", + " 15626229210687624776\n", + " 63.0\n", " male\n", " white_not_hispanic_or_latino\n", " employed\n", " in_person\n", - " 297058.0\n", + " 7224.0\n", " private_auto\n", " 2.0\n", - " 380443.0\n", + " 23805.0\n", " two\n", " core\n", - " naics62\n", + " naics713940\n", " multiple_units\n", " not_attending_school\n", - " bachelors_degree\n", - " owner\n", - " english\n", - " 1 (Tract 2643.05, Los Angeles, CA)\n", - " 2643.05 (Los Angeles, CA)\n", + " some_college\n", + " renter\n", + " indo_european\n", + " 1 (Tract 1349.07, Los Angeles, CA)\n", + " 1349.07 (Los Angeles, CA)\n", " Los Angeles County, CA\n", " California\n", - " 3 (Tract 9105.01, Los Angeles, CA)\n", - " 9105.01 (Los Angeles, CA)\n", + " 3 (Tract 7004, Los Angeles, CA)\n", + " 7004 (Los Angeles, CA)\n", " Los Angeles County, CA\n", " California\n", " \n", " \n", - " 245631\n", - " 12533923940243185590\n", - " 2 (Tract 6203.03, Los Angeles, CA)\n", - " 6203.03 (Los Angeles, CA)\n", + " 101372\n", + " 12431969978777812348\n", + " 1 (Tract 1174.08, Los Angeles, CA)\n", + " 1174.08 (Los Angeles, CA)\n", " Los Angeles County, CA\n", " California\n", - " 1 (Tract 29.14, Santa Barbara, CA)\n", - " 29.14 (Santa Barbara, CA)\n", - " Santa Barbara County, CA\n", + " 3 (Tract 2735.02, Los Angeles, CA)\n", + " 2735.02 (Los Angeles, CA)\n", + " Los Angeles County, CA\n", " California\n", " private_auto\n", - " home\n", " shop\n", - " 17:30:40\n", - " 19:53:16\n", - " 142\n", - " 110.3\n", + " home\n", + " 17:16:00\n", + " 18:01:55\n", + " 45\n", + " 20.4\n", " unknown_vehicle_type\n", " other_non_bev\n", " NaN\n", " NaN\n", " NaN\n", + " single_family\n", + " single_family\n", " retail\n", " retail\n", - " multi_family\n", - " multi_family\n", - " 15212051208025649622\n", - " 13620258903763937995\n", - " 69.0\n", - " male\n", - " white_not_hispanic_or_latino\n", + " 3550536633128717092\n", + " 6516243422276820403\n", + " 62.0\n", + " female\n", + " asian_not_hispanic_or_latino\n", " employed\n", " in_person\n", - " 46394.0\n", + " 27836.0\n", " private_auto\n", - " 1.0\n", - " 46394.0\n", - " one\n", + " 4.0\n", + " 110187.0\n", + " two\n", " core\n", - " naics515210\n", - " multiple_units\n", - " not_attending_school\n", + " naics621111\n", + " single_family\n", + " graduate\n", " bachelors_degree\n", - " renter\n", - " english\n", - " 1 (Tract 29.14, Santa Barbara, CA)\n", - " 29.14 (Santa Barbara, CA)\n", - " Santa Barbara County, CA\n", + " owner\n", + " other\n", + " 1 (Tract 1174.08, Los Angeles, CA)\n", + " 1174.08 (Los Angeles, CA)\n", + " Los Angeles County, CA\n", " California\n", - " 1 (Tract 6204, Los Angeles, CA)\n", - " 6204 (Los Angeles, CA)\n", + " 3 (Tract 5435.03, Los Angeles, CA)\n", + " 5435.03 (Los Angeles, CA)\n", " Los Angeles County, CA\n", " California\n", " \n", " \n", - " 258108\n", - " 15822036852902980413\n", - " 1 (Tract 2643.06, Los Angeles, CA)\n", - " 2643.06 (Los Angeles, CA)\n", + " 237178\n", + " 1520438980495662157\n", + " 4 (Tract 1175.30, Los Angeles, CA)\n", + " 1175.30 (Los Angeles, CA)\n", " Los Angeles County, CA\n", " California\n", - " 1 (Tract 1043.21, Los Angeles, CA)\n", - " 1043.21 (Los Angeles, CA)\n", + " 1 (Tract 9800.28, Los Angeles, CA)\n", + " 9800.28 (Los Angeles, CA)\n", " Los Angeles County, CA\n", " California\n", " private_auto\n", - " home\n", " work\n", - " 14:52:00\n", - " 15:40:55\n", - " 48\n", - " 19.9\n", + " home\n", + " 07:19:00\n", + " 08:13:36\n", + " 54\n", + " 23.1\n", " unknown_vehicle_type\n", " other_non_bev\n", " NaN\n", " NaN\n", " NaN\n", - " retail\n", - " retail\n", " multi_family\n", " multi_family\n", - " 18024054114643477512\n", - " 4390175249257531498\n", - " 32.0\n", + " industrial\n", + " industrial\n", + " 8591229506040203519\n", + " 16428891936194534442\n", + " 24.0\n", " male\n", " hispanic_or_latino_origin\n", " employed\n", " in_person\n", - " 12507.0\n", + " 40595.0\n", " private_auto\n", - " 2.0\n", - " 56284.0\n", - " two\n", + " 4.0\n", + " 102068.0\n", + " three_plus\n", " core\n", - " naics23\n", + " naics44_45\n", " multiple_units\n", " not_attending_school\n", - " k_12\n", + " some_college\n", " renter\n", " spanish\n", - " 1 (Tract 1043.21, Los Angeles, CA)\n", - " 1043.21 (Los Angeles, CA)\n", + " 4 (Tract 1175.30, Los Angeles, CA)\n", + " 1175.30 (Los Angeles, CA)\n", " Los Angeles County, CA\n", " California\n", - " 1 (Tract 2643.06, Los Angeles, CA)\n", - " 2643.06 (Los Angeles, CA)\n", + " 1 (Tract 9800.28, Los Angeles, CA)\n", + " 9800.28 (Los Angeles, CA)\n", " Los Angeles County, CA\n", " California\n", " \n", " \n", - " 146285\n", - " 14712612269617851560\n", - " 1 (Tract 1112.01, Los Angeles, CA)\n", - " 1112.01 (Los Angeles, CA)\n", + " 31761\n", + " 11036673894851484580\n", + " 1 (Tract 6506.06, Los Angeles, CA)\n", + " 6506.06 (Los Angeles, CA)\n", " Los Angeles County, CA\n", " California\n", - " 1 (Tract 2653.01, Los Angeles, CA)\n", - " 2653.01 (Los Angeles, CA)\n", - " Los Angeles County, CA\n", + " 2 (Tract 56, Kern, CA)\n", + " 56 (Kern, CA)\n", + " Kern County, CA\n", " California\n", " private_auto\n", - " school\n", - " eat\n", - " 07:59:00\n", - " 08:49:01\n", - " 50\n", - " 17.9\n", + " work\n", + " social\n", + " 07:00:00\n", + " 09:53:47\n", + " 173\n", + " 142.9\n", " unknown_vehicle_type\n", " other_non_bev\n", " NaN\n", " NaN\n", " NaN\n", + " multi_family\n", + " multi_family\n", " retail\n", " retail\n", - " education\n", - " education\n", - " 6730005140840342218\n", - " 8465342313702012328\n", - " 20.0\n", - " male\n", - " hispanic_or_latino_origin\n", - " not_in_labor_force\n", - " unemployed_under_16_not_in_labor_force\n", - " 0.0\n", - " other_travel_mode\n", - " 3.0\n", - " 5211.0\n", - " two\n", + " 673878670446905249\n", + " 1964084087473365831\n", + " 42.0\n", + " female\n", + " asian_not_hispanic_or_latino\n", + " employed\n", + " in_person\n", + " 23942.0\n", + " private_auto\n", + " 2.0\n", + " 80947.0\n", + " one\n", " core\n", - " not_working\n", + " naics23\n", " multiple_units\n", - " undergraduate\n", - " some_college\n", + " not_attending_school\n", + " bachelors_degree\n", " renter\n", - " spanish\n", - " 1 (Tract 2653.04, Los Angeles, CA)\n", - " 2653.04 (Los Angeles, CA)\n", + " asian_pacific\n", + " 1 (Tract 6506.06, Los Angeles, CA)\n", + " 6506.06 (Los Angeles, CA)\n", " Los Angeles County, CA\n", " California\n", - " 1 (Tract 2653.01, Los Angeles, CA)\n", - " 2653.01 (Los Angeles, CA)\n", - " Los Angeles County, CA\n", + " 2 (Tract 56, Kern, CA)\n", + " 56 (Kern, CA)\n", + " Kern County, CA\n", " California\n", " \n", " \n", - " 58235\n", - " 16619945285298383268\n", - " 1 (Tract 5436.07, Los Angeles, CA)\n", - " 5436.07 (Los Angeles, CA)\n", + " 193932\n", + " 7236618342537072261\n", + " 1 (Tract 2733, Los Angeles, CA)\n", + " 2733 (Los Angeles, CA)\n", " Los Angeles County, CA\n", " California\n", - " 1 (Tract 1321.02, Los Angeles, CA)\n", - " 1321.02 (Los Angeles, CA)\n", + " 1 (Tract 9800.08, Los Angeles, CA)\n", + " 9800.08 (Los Angeles, CA)\n", " Los Angeles County, CA\n", " California\n", - " private_auto\n", + " auto_passenger\n", " work\n", - " home\n", - " 12:39:00\n", - " 13:28:17\n", - " 49\n", - " 34.9\n", + " eat\n", + " 09:19:07\n", + " 09:46:59\n", + " 27\n", + " 20.0\n", " unknown_vehicle_type\n", - " other_non_bev\n", + " unknown_fuel_type\n", " NaN\n", " NaN\n", " NaN\n", - " single_family\n", - " single_family\n", " retail\n", " retail\n", - " 14684325219070757087\n", - " 7566196984556086662\n", - " 64.0\n", - " male\n", + " mixed_use\n", + " office\n", + " 2321481644760744007\n", + " 11390460785703966067\n", + " 24.0\n", + " female\n", " asian_not_hispanic_or_latino\n", " employed\n", " in_person\n", - " 83858.0\n", - " private_auto\n", - " 1.0\n", - " 83858.0\n", - " one\n", + " 27831.0\n", + " public_transit\n", + " 9.0\n", + " 256361.0\n", + " three_plus\n", " core\n", - " naics44_45\n", - " several_units\n", + " naics621493\n", + " single_family\n", " not_attending_school\n", - " some_college\n", + " bachelors_degree\n", " owner\n", " asian_pacific\n", - " 1 (Tract 5436.07, Los Angeles, CA)\n", - " 5436.07 (Los Angeles, CA)\n", + " 2 (Tract 1958.03, Los Angeles, CA)\n", + " 1958.03 (Los Angeles, CA)\n", " Los Angeles County, CA\n", " California\n", - " 1 (Tract 1321.02, Los Angeles, CA)\n", - " 1321.02 (Los Angeles, CA)\n", + " 1 (Tract 9800.08, Los Angeles, CA)\n", + " 9800.08 (Los Angeles, CA)\n", " Los Angeles County, CA\n", " California\n", " \n", @@ -503,151 +503,151 @@ ], "text/plain": [ " activity_id origin_bgrp_2020 \\\n", - "423350 5926084662673022461 1 (Tract 2643.05, Los Angeles, CA) \n", - "245631 12533923940243185590 2 (Tract 6203.03, Los Angeles, CA) \n", - "258108 15822036852902980413 1 (Tract 2643.06, Los Angeles, CA) \n", - "146285 14712612269617851560 1 (Tract 1112.01, Los Angeles, CA) \n", - "58235 16619945285298383268 1 (Tract 5436.07, Los Angeles, CA) \n", + "314707 8505387565750628825 1 (Tract 1349.07, Los Angeles, CA) \n", + "101372 12431969978777812348 1 (Tract 1174.08, Los Angeles, CA) \n", + "237178 1520438980495662157 4 (Tract 1175.30, Los Angeles, CA) \n", + "31761 11036673894851484580 1 (Tract 6506.06, Los Angeles, CA) \n", + "193932 7236618342537072261 1 (Tract 2733, Los Angeles, CA) \n", "\n", " origin_trct_2020 origin_cty_2020 origin_st_2020 \\\n", - "423350 2643.05 (Los Angeles, CA) Los Angeles County, CA California \n", - "245631 6203.03 (Los Angeles, CA) Los Angeles County, CA California \n", - "258108 2643.06 (Los Angeles, CA) Los Angeles County, CA California \n", - "146285 1112.01 (Los Angeles, CA) Los Angeles County, CA California \n", - "58235 5436.07 (Los Angeles, CA) Los Angeles County, CA California \n", + "314707 1349.07 (Los Angeles, CA) Los Angeles County, CA California \n", + "101372 1174.08 (Los Angeles, CA) Los Angeles County, CA California \n", + "237178 1175.30 (Los Angeles, CA) Los Angeles County, CA California \n", + "31761 6506.06 (Los Angeles, CA) Los Angeles County, CA California \n", + "193932 2733 (Los Angeles, CA) Los Angeles County, CA California \n", "\n", " destination_bgrp_2020 destination_trct_2020 \\\n", - "423350 3 (Tract 9105.01, Los Angeles, CA) 9105.01 (Los Angeles, CA) \n", - "245631 1 (Tract 29.14, Santa Barbara, CA) 29.14 (Santa Barbara, CA) \n", - "258108 1 (Tract 1043.21, Los Angeles, CA) 1043.21 (Los Angeles, CA) \n", - "146285 1 (Tract 2653.01, Los Angeles, CA) 2653.01 (Los Angeles, CA) \n", - "58235 1 (Tract 1321.02, Los Angeles, CA) 1321.02 (Los Angeles, CA) \n", - "\n", - " destination_cty_2020 destination_st_2020 primary_mode \\\n", - "423350 Los Angeles County, CA California private_auto \n", - "245631 Santa Barbara County, CA California private_auto \n", - "258108 Los Angeles County, CA California private_auto \n", - "146285 Los Angeles County, CA California private_auto \n", - "58235 Los Angeles County, CA California private_auto \n", + "314707 4 (Tract 6037.02, Los Angeles, CA) 6037.02 (Los Angeles, CA) \n", + "101372 3 (Tract 2735.02, Los Angeles, CA) 2735.02 (Los Angeles, CA) \n", + "237178 1 (Tract 9800.28, Los Angeles, CA) 9800.28 (Los Angeles, CA) \n", + "31761 2 (Tract 56, Kern, CA) 56 (Kern, CA) \n", + "193932 1 (Tract 9800.08, Los Angeles, CA) 9800.08 (Los Angeles, CA) \n", + "\n", + " destination_cty_2020 destination_st_2020 primary_mode \\\n", + "314707 Los Angeles County, CA California private_auto \n", + "101372 Los Angeles County, CA California private_auto \n", + "237178 Los Angeles County, CA California private_auto \n", + "31761 Kern County, CA California private_auto \n", + "193932 Los Angeles County, CA California auto_passenger \n", "\n", " trip_purpose previous_trip_purpose trip_start_time trip_end_time \\\n", - "423350 work home 14:57:00 16:25:56 \n", - "245631 home shop 17:30:40 19:53:16 \n", - "258108 home work 14:52:00 15:40:55 \n", - "146285 school eat 07:59:00 08:49:01 \n", - "58235 work home 12:39:00 13:28:17 \n", + "314707 social home 13:31:00 14:24:59 \n", + "101372 shop home 17:16:00 18:01:55 \n", + "237178 work home 07:19:00 08:13:36 \n", + "31761 work social 07:00:00 09:53:47 \n", + "193932 work eat 09:19:07 09:46:59 \n", "\n", " trip_duration_minutes trip_distance_miles vehicle_type \\\n", - "423350 88 57.2 unknown_vehicle_type \n", - "245631 142 110.3 unknown_vehicle_type \n", - "258108 48 19.9 unknown_vehicle_type \n", - "146285 50 17.9 unknown_vehicle_type \n", - "58235 49 34.9 unknown_vehicle_type \n", - "\n", - " vehicle_fuel_type transit_submode transit_agency transit_route \\\n", - "423350 other_non_bev NaN NaN NaN \n", - "245631 other_non_bev NaN NaN NaN \n", - "258108 other_non_bev NaN NaN NaN \n", - "146285 other_non_bev NaN NaN NaN \n", - "58235 other_non_bev NaN NaN NaN \n", + "314707 53 31.2 unknown_vehicle_type \n", + "101372 45 20.4 unknown_vehicle_type \n", + "237178 54 23.1 unknown_vehicle_type \n", + "31761 173 142.9 unknown_vehicle_type \n", + "193932 27 20.0 unknown_vehicle_type \n", + "\n", + " vehicle_fuel_type transit_submode transit_agency transit_route \\\n", + "314707 other_non_bev NaN NaN NaN \n", + "101372 other_non_bev NaN NaN NaN \n", + "237178 other_non_bev NaN NaN NaN \n", + "31761 other_non_bev NaN NaN NaN \n", + "193932 unknown_fuel_type NaN NaN NaN \n", "\n", " origin_land_use origin_building_use destination_land_use \\\n", - "423350 mixed_use multi_family civic_institutional \n", - "245631 retail retail multi_family \n", - "258108 retail retail multi_family \n", - "146285 retail retail education \n", - "58235 single_family single_family retail \n", + "314707 mixed_use multi_family single_family \n", + "101372 single_family single_family retail \n", + "237178 multi_family multi_family industrial \n", + "31761 multi_family multi_family retail \n", + "193932 retail retail mixed_use \n", "\n", " destination_building_use trip_taker_person_id trip_taker_household_id \\\n", - "423350 civic_institutional 9280162199592818201 3819422246701548245 \n", - "245631 multi_family 15212051208025649622 13620258903763937995 \n", - "258108 multi_family 18024054114643477512 4390175249257531498 \n", - "146285 education 6730005140840342218 8465342313702012328 \n", - "58235 retail 14684325219070757087 7566196984556086662 \n", + "314707 single_family 13590816709645713585 15626229210687624776 \n", + "101372 retail 3550536633128717092 6516243422276820403 \n", + "237178 industrial 8591229506040203519 16428891936194534442 \n", + "31761 retail 673878670446905249 1964084087473365831 \n", + "193932 office 2321481644760744007 11390460785703966067 \n", "\n", " trip_taker_age trip_taker_sex trip_taker_race_ethnicity \\\n", - "423350 59.0 male white_not_hispanic_or_latino \n", - "245631 69.0 male white_not_hispanic_or_latino \n", - "258108 32.0 male hispanic_or_latino_origin \n", - "146285 20.0 male hispanic_or_latino_origin \n", - "58235 64.0 male asian_not_hispanic_or_latino \n", - "\n", - " trip_taker_employment_status trip_taker_wfh \\\n", - "423350 employed in_person \n", - "245631 employed in_person \n", - "258108 employed in_person \n", - "146285 not_in_labor_force unemployed_under_16_not_in_labor_force \n", - "58235 employed in_person \n", + "314707 63.0 male white_not_hispanic_or_latino \n", + "101372 62.0 female asian_not_hispanic_or_latino \n", + "237178 24.0 male hispanic_or_latino_origin \n", + "31761 42.0 female asian_not_hispanic_or_latino \n", + "193932 24.0 female asian_not_hispanic_or_latino \n", + "\n", + " trip_taker_employment_status trip_taker_wfh \\\n", + "314707 employed in_person \n", + "101372 employed in_person \n", + "237178 employed in_person \n", + "31761 employed in_person \n", + "193932 employed in_person \n", "\n", " trip_taker_individual_income trip_taker_commute_mode \\\n", - "423350 297058.0 private_auto \n", - "245631 46394.0 private_auto \n", - "258108 12507.0 private_auto \n", - "146285 0.0 other_travel_mode \n", - "58235 83858.0 private_auto \n", + "314707 7224.0 private_auto \n", + "101372 27836.0 private_auto \n", + "237178 40595.0 private_auto \n", + "31761 23942.0 private_auto \n", + "193932 27831.0 public_transit \n", "\n", " trip_taker_household_size trip_taker_household_income \\\n", - "423350 2.0 380443.0 \n", - "245631 1.0 46394.0 \n", - "258108 2.0 56284.0 \n", - "146285 3.0 5211.0 \n", - "58235 1.0 83858.0 \n", + "314707 2.0 23805.0 \n", + "101372 4.0 110187.0 \n", + "237178 4.0 102068.0 \n", + "31761 2.0 80947.0 \n", + "193932 9.0 256361.0 \n", "\n", " trip_taker_available_vehicles trip_taker_resident_type \\\n", - "423350 two core \n", - "245631 one core \n", - "258108 two core \n", - "146285 two core \n", - "58235 one core \n", + "314707 two core \n", + "101372 two core \n", + "237178 three_plus core \n", + "31761 one core \n", + "193932 three_plus core \n", "\n", " trip_taker_industry trip_taker_building_type \\\n", - "423350 naics62 multiple_units \n", - "245631 naics515210 multiple_units \n", - "258108 naics23 multiple_units \n", - "146285 not_working multiple_units \n", - "58235 naics44_45 several_units \n", + "314707 naics713940 multiple_units \n", + "101372 naics621111 single_family \n", + "237178 naics44_45 multiple_units \n", + "31761 naics23 multiple_units \n", + "193932 naics621493 single_family \n", "\n", " trip_taker_school_grade_attending trip_taker_education \\\n", - "423350 not_attending_school bachelors_degree \n", - "245631 not_attending_school bachelors_degree \n", - "258108 not_attending_school k_12 \n", - "146285 undergraduate some_college \n", - "58235 not_attending_school some_college \n", + "314707 not_attending_school some_college \n", + "101372 graduate bachelors_degree \n", + "237178 not_attending_school some_college \n", + "31761 not_attending_school bachelors_degree \n", + "193932 not_attending_school bachelors_degree \n", "\n", " trip_taker_tenure trip_taker_language \\\n", - "423350 owner english \n", - "245631 renter english \n", - "258108 renter spanish \n", - "146285 renter spanish \n", - "58235 owner asian_pacific \n", + "314707 renter indo_european \n", + "101372 owner other \n", + "237178 renter spanish \n", + "31761 renter asian_pacific \n", + "193932 owner asian_pacific \n", "\n", " trip_taker_home_bgrp_2020 trip_taker_home_trct_2020 \\\n", - "423350 1 (Tract 2643.05, Los Angeles, CA) 2643.05 (Los Angeles, CA) \n", - "245631 1 (Tract 29.14, Santa Barbara, CA) 29.14 (Santa Barbara, CA) \n", - "258108 1 (Tract 1043.21, Los Angeles, CA) 1043.21 (Los Angeles, CA) \n", - "146285 1 (Tract 2653.04, Los Angeles, CA) 2653.04 (Los Angeles, CA) \n", - "58235 1 (Tract 5436.07, Los Angeles, CA) 5436.07 (Los Angeles, CA) \n", - "\n", - " trip_taker_home_cty_2020 trip_taker_home_st_2020 \\\n", - "423350 Los Angeles County, CA California \n", - "245631 Santa Barbara County, CA California \n", - "258108 Los Angeles County, CA California \n", - "146285 Los Angeles County, CA California \n", - "58235 Los Angeles County, CA California \n", + "314707 1 (Tract 1349.07, Los Angeles, CA) 1349.07 (Los Angeles, CA) \n", + "101372 1 (Tract 1174.08, Los Angeles, CA) 1174.08 (Los Angeles, CA) \n", + "237178 4 (Tract 1175.30, Los Angeles, CA) 1175.30 (Los Angeles, CA) \n", + "31761 1 (Tract 6506.06, Los Angeles, CA) 6506.06 (Los Angeles, CA) \n", + "193932 2 (Tract 1958.03, Los Angeles, CA) 1958.03 (Los Angeles, CA) \n", + "\n", + " trip_taker_home_cty_2020 trip_taker_home_st_2020 \\\n", + "314707 Los Angeles County, CA California \n", + "101372 Los Angeles County, CA California \n", + "237178 Los Angeles County, CA California \n", + "31761 Los Angeles County, CA California \n", + "193932 Los Angeles County, CA California \n", "\n", " trip_taker_work_bgrp_2020 trip_taker_work_trct_2020 \\\n", - "423350 3 (Tract 9105.01, Los Angeles, CA) 9105.01 (Los Angeles, CA) \n", - "245631 1 (Tract 6204, Los Angeles, CA) 6204 (Los Angeles, CA) \n", - "258108 1 (Tract 2643.06, Los Angeles, CA) 2643.06 (Los Angeles, CA) \n", - "146285 1 (Tract 2653.01, Los Angeles, CA) 2653.01 (Los Angeles, CA) \n", - "58235 1 (Tract 1321.02, Los Angeles, CA) 1321.02 (Los Angeles, CA) \n", + "314707 3 (Tract 7004, Los Angeles, CA) 7004 (Los Angeles, CA) \n", + "101372 3 (Tract 5435.03, Los Angeles, CA) 5435.03 (Los Angeles, CA) \n", + "237178 1 (Tract 9800.28, Los Angeles, CA) 9800.28 (Los Angeles, CA) \n", + "31761 2 (Tract 56, Kern, CA) 56 (Kern, CA) \n", + "193932 1 (Tract 9800.08, Los Angeles, CA) 9800.08 (Los Angeles, CA) \n", "\n", " trip_taker_work_cty_2020 trip_taker_work_st_2020 \n", - "423350 Los Angeles County, CA California \n", - "245631 Los Angeles County, CA California \n", - "258108 Los Angeles County, CA California \n", - "146285 Los Angeles County, CA California \n", - "58235 Los Angeles County, CA California " + "314707 Los Angeles County, CA California \n", + "101372 Los Angeles County, CA California \n", + "237178 Los Angeles County, CA California \n", + "31761 Kern County, CA California \n", + "193932 Los Angeles County, CA California " ] }, "execution_count": 8, @@ -788,23 +788,23 @@ "text/html": [ "\n", "\n", - "
\n", + "
\n", "" ], "text/plain": [ "alt.Chart(...)" ] }, - "execution_count": 55, + "execution_count": 53, "metadata": {}, "output_type": "execute_result" } @@ -3046,7 +2833,7 @@ }, { "cell_type": "code", - "execution_count": 56, + "execution_count": 54, "id": "60df338f-801f-49a1-a26f-0be5e03bddf3", "metadata": {}, "outputs": [ @@ -3120,7 +2907,7 @@ "1 Metro - Los Angeles" ] }, - "execution_count": 56, + "execution_count": 54, "metadata": {}, "output_type": "execute_result" } @@ -3149,102 +2936,179 @@ }, { "cell_type": "code", - "execution_count": 57, - "id": "8af03d60-4537-44d3-b82f-d857625e47c3", - "metadata": {}, - "outputs": [], - "source": [ - "columns_to_keep = [\"AVTA\",\n", - " \"Beach Cities Transit-City of Redondo Beach\",\n", - " \"LADOT126\",\n", - " \"Metro - Los Angeles\",\n", - " \"LA Go Bus\",\n", - " \"LADOTDT\",\n", - " \"Amtrak\",\n", - " \"Big Blue Bus\",\n", - " \"GTrans\",\n", - " \"LADOTMVN\",\n", - " \"Long Beach Transit\",\n", - " \"Santa Clarita Transit\",\n", - " \"Foothill Transit\",\n", - " \"Culver CityBus\",\n", - " \"Metrolink Trains\",\n", - " \"LADOTMVS\"\n", - " ]" - ] - }, - { - "cell_type": "code", - "execution_count": 58, - "id": "6951b590-cef8-44bd-b27d-0b18893609c5", - "metadata": {}, - "outputs": [], - "source": [ - "# agency_indv_count.sample()" - ] - }, - { - "cell_type": "code", - "execution_count": 59, - "id": "cca56d58-5e3f-430c-adde-15ca665d46f8", - "metadata": {}, - "outputs": [], - "source": [ - "df_agencies = agency_indv_count.loc[:, columns_to_keep]" - ] - }, - { - "cell_type": "code", - "execution_count": 60, - "id": "814dde3a-d303-4044-9df1-464e408bf460", - "metadata": {}, - "outputs": [], - "source": [ - "df_agencies = pd.DataFrame(df_agencies).transpose().reset_index().rename(columns={'index':'agency'})" - ] - }, - { - "cell_type": "code", - "execution_count": 61, - "id": "84174911-2243-430c-ab73-1fecb2552094", + "execution_count": null, + "id": "febd910a-2f1d-4b99-bb3b-76adb11f0b63", "metadata": {}, "outputs": [], - "source": [ - "# df_agencies" - ] + "source": [] }, { "cell_type": "code", - "execution_count": 62, - "id": "202923d2-b869-4a72-8a02-ce229a65d815", + "execution_count": 55, + "id": "96b794f2-bcfd-4afb-9ce5-8793c15fe199", "metadata": {}, "outputs": [ { "name": "stderr", "output_type": "stream", "text": [ - "/tmp/ipykernel_515/2517619377.py:1: FutureWarning: Dropping of nuisance columns in DataFrame reductions (with 'numeric_only=None') is deprecated; in a future version this will raise TypeError. Select only valid columns before calling the reduction.\n", + "/home/jovyan/data-analyses/sb125_analyses/corridor_study/_utils.py:98: FutureWarning: Dropping of nuisance columns in DataFrame reductions (with 'numeric_only=None') is deprecated; in a future version this will raise TypeError. Select only valid columns before calling the reduction.\n", " df_agencies['n_trips'] = df_agencies[list(df_agencies.columns)].sum(axis=1)\n" ] } ], "source": [ - "df_agencies['n_trips'] = df_agencies[list(df_agencies.columns)].sum(axis=1)" + "df_agencies = _utils.get_agencies_occurances(df)" ] }, { "cell_type": "code", - "execution_count": 63, - "id": "2b31eb23-e1e6-43c6-8b0e-ccb3802cd09d", + "execution_count": 56, + "id": "80cbffde-6686-4921-b543-84e52f97a0c5", "metadata": {}, - "outputs": [], + "outputs": [ + { + "data": { + "text/html": [ + "
\n", + "\n", + "\n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + "
agencyn_trips
0Culver CityBus28
1Foothill Transit1
2LA Go Bus1
3Metro - Los Angeles195
4Amtrak8
5AVTA4
6GTrans1
7Metrolink Trains1
8LADOTMVN54
9Long Beach Transit1
10Santa Clarita Transit28
11LADOTDT3
12Big Blue Bus37
13LADOT1264
14LADOTMVS4
15Beach Cities Transit-City of Redondo Beach2
\n", + "
" + ], + "text/plain": [ + " agency n_trips\n", + "0 Culver CityBus 28\n", + "1 Foothill Transit 1\n", + "2 LA Go Bus 1\n", + "3 Metro - Los Angeles 195\n", + "4 Amtrak 8\n", + "5 AVTA 4\n", + "6 GTrans 1\n", + "7 Metrolink Trains 1\n", + "8 LADOTMVN 54\n", + "9 Long Beach Transit 1\n", + "10 Santa Clarita Transit 28\n", + "11 LADOTDT 3\n", + "12 Big Blue Bus 37\n", + "13 LADOT126 4\n", + "14 LADOTMVS 4\n", + "15 Beach Cities Transit-City of Redondo Beach 2" + ] + }, + "execution_count": 56, + "metadata": {}, + "output_type": "execute_result" + } + ], "source": [ - "df_agencies = df_agencies>>select(_.agency, _.n_trips)" + "df_agencies" ] }, { "cell_type": "code", - "execution_count": 64, + "execution_count": 57, "id": "f98c76b9-b83f-42c3-aa12-702ade72ac38", "metadata": {}, "outputs": [], @@ -3254,17 +3118,7 @@ }, { "cell_type": "code", - "execution_count": 65, - "id": "f1267faf-e28c-4208-ae79-43b31f67f705", - "metadata": {}, - "outputs": [], - "source": [ - "# df_agencies" - ] - }, - { - "cell_type": "code", - "execution_count": 66, + "execution_count": 59, "id": "bf160b81-1bc2-4374-9529-6400252d5e4f", "metadata": {}, "outputs": [ @@ -3273,23 +3127,23 @@ "text/html": [ "\n", "\n", - "
\n", + "
\n", "" ], "text/plain": [ "alt.Chart(...)" ] }, - "execution_count": 66, + "execution_count": 59, "metadata": {}, "output_type": "execute_result" } @@ -3372,7 +3226,7 @@ }, { "cell_type": "code", - "execution_count": 67, + "execution_count": 60, "id": "16b45917-adc3-44d0-8d76-15ccd4083d26", "metadata": {}, "outputs": [ @@ -3395,7 +3249,7 @@ }, { "cell_type": "code", - "execution_count": 68, + "execution_count": 61, "id": "5163926e-82bd-4c25-9486-45229d09a0d1", "metadata": {}, "outputs": [], @@ -3408,7 +3262,7 @@ }, { "cell_type": "code", - "execution_count": 69, + "execution_count": 62, "id": "636e7771-0ba7-42e9-af1d-e3e998ff8599", "metadata": {}, "outputs": [ @@ -3531,7 +3385,7 @@ "11 public_transit visitor 56" ] }, - "execution_count": 69, + "execution_count": 62, "metadata": {}, "output_type": "execute_result" } @@ -3542,7 +3396,7 @@ }, { "cell_type": "code", - "execution_count": 70, + "execution_count": 63, "id": "90c5d683-91c5-4663-accd-4e45027a3e49", "metadata": {}, "outputs": [ @@ -3551,23 +3405,23 @@ "text/html": [ "\n", "\n", - "
\n", + "
\n", "