d4c-datapkg-statistical/experiments/presentation/cop_2021_structure_example.ipynb

In [1]:

import duckdb
import pandas as pd

In [2]:

con = duckdb.connect()
con.install_extension("spatial")
con.load_extension("spatial")

In [3]:

con.execute("""
DROP TABLE IF EXISTS geo_data;
CREATE TABLE geo_data AS
SELECT * FROM 'https://data-01.dataforcanada.org/processed/statistics_canada/census_of_population/2021/tabular/da_2021.parquet';
""")

Out [3]:

FloatProgress(value=0.0, layout=Layout(width='auto'), style=ProgressStyle(bar_color='black'))

<duckdb.duckdb.DuckDBPyConnection at 0x7fd222a8fb70>

In [4]:

geo_data = con.sql("SELECT * FROM geo_data").to_df()

In [5]:

geo_data

Out [5]:

	da_dguid	count_total_1	count_total_4	count_total_5	count_total_6	count_total_7	count_total_8	count_total_9	count_total_10	count_total_11	...	count_women_2622	count_women_2623	count_women_2624	count_women_2625	count_women_2626	count_women_2627	count_women_2628	count_women_2629	count_women_2630	count_women_2631
0	2021S051210010165	500	229	217	1309.9	0.38	500	45	10	20	...	0	15	25	0	25	20	0	0	0	20
1	2021S051210010166	340	165	146	3563.9	0.10	340	50	15	25	...	15	0	30	0	30	20	0	0	0	20
2	2021S051210010167	407	201	184	3217.4	0.13	405	45	15	10	...	10	0	35	0	35	25	0	0	0	25
3	2021S051210010168	523	274	248	2736.8	0.19	525	55	15	15	...	15	30	35	0	35	25	0	0	0	20
4	2021S051210010169	285	131	123	4305.1	0.07	285	40	10	15	...	10	10	20	0	20	15	0	0	0	15
...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...
57931	2021S051262080024	423	166	140	1055.9	0.40	425	90	25	35	...	15	10	55	0	50	40	0	0	0	40
57932	2021S051262080025	514	162	148	1.8	287.39	515	160	50	55	...	15	10	90	5	90	65	0	0	0	60
57933	2021S051262080026	868	276	249	3.4	251.60	865	285	100	85	...	20	15	170	10	160	120	5	5	0	110
57934	2021S051262080027	776	196	169	31.5	24.65	775	300	115	90	...	20	10	160	0	160	100	0	0	0	100
57935	2021S051262080028	573	143	123	146.9	3.90	575	205	75	75	...	10	15	105	5	105	70	0	0	0	65

57936 rows × 7432 columns

In [ ]: