snap-stanford/covid-mobility

covid-mobility's Issues

Source for 2018 1-year pop estimates at block group level?

I am unable to find such a dataset for geographies smaller than tract. Could you point me to your source?
Thanks,
~~ Scott

Suggestion / request to make data navigation falling-off-a-tree simple

I used your constants and utils to create the structure below. I imagine I can eventually suss out source data locations on the tree given your very good documentation. It would help, however, if you could populate such a tree with those file names

├── scratch1
│   └── safegraph_homes
│   ├── all_aggregate_data
│   │   ├── 20191213-safegraph-aggregate-longitudinal-data-to-unzip-to
│   │   │   └── SearchofAllRecords-CORE_POI-GEOMETRY-PATTERNS-2020_02-2020-03-16
│   │   │   └── visit_panel_summary.csv
│   │   ├── chunks_with_demographic_annotations
│   │   ├── chunks_with_demographic_annotations_stratified_by_area
│   │   ├── daily_counts_of_people_leaving_homes
│   │   │   ├── sg-social-distancing
│   │   │   └── social_distancing_v2
│   │   ├── fitted_models
│   │   ├── ipf_output
│   │   ├── safegraph_poi_area_calculations
│   │   │   └── SafeGraphPlacesGeoSupplementSquareFeet.csv.gz
│   │   └── weekly_patterns_data
│   │   └── v1
│   │   ├── home_summary_file
│   │   └── main-file
│   ├── base_dir_for_all_new_data_and_results
│   │   └── non_safegraph_datasets
│   │   └── census_block_group_data
│   │   ├── ACS_5_year_2013_to_2017_joined_to_blockgroup_shapefiles
│   │   └── august_2017_county_to_metropolitan_mapping.csv
│   ├── external_datasets_for_aggregate_analysis
│   │   ├── census_block_group_shapefiles_by_state
│   │   └── msa_shapefiles
│   │   └── tl_2017_us_cbsa
│   └── old_dfs_scratch0_directory_contents
│   └── new_census_data
└── scratch2
└── second_safegraph_homes
└── extra_safegraph_aggregate_models

Varied solutions and mismatched row-columns for "Iterative Proportional Fitting Procedure"

Hi, thanks for this amazing paper, it's very interesting and insightful. I have two questions about the dataset, and I was wondering if anyone could help me take a look when available.

1. The Iterative Proportional Fitting Procedure always gives me varied solutions, whenever I change the initial point, the final results will change. It makes sense since we are trying to decompose weekly data into hourly data, there are always multiple possible solutions, but how do you handle this problem in this paper? Which solution will you use because it is very important for the final results?
2. Mismatch between the sum of rows and the sum of columns in the Iterative Proportional Fitting Procedure. For example, from the time point of view, 1,3,2,4,5 people visit this POI (I only count 5 hours for simplify), so we have 15 visitors in total today. But from CBG point of view, we have 5 people from CBG1,10 people from CBG2 and 20 people from CBG3. So we have 35 visitor in total. How to overcome this mismatch?

Thank you!!!

Referencing file paths

Would it be possible to add a section in the README regarding which part of the codes need to be changed in order to properly specify file paths? At first, I thought it was only the covid_constants_and_utils.py file. But, I saw some other files make references to filepaths as well.

oo

oo