Duplicate ID across different excel files

[choisy]

IDs 4445, 4536, 4582, 4600, 4632, 4831, 4840, 4897, 5018, 5019, 5033, 5122, 5217, 5218, 5251, 5252, 5254, 6323, 7413, 7436, 8101 are on raw_data/GPS/Shared with app.xls and raw_data/GPS/New Machine2015-2018.xls files at the same time.

ID 8101 is both on raw_data/GPS/IPL Server.xlsx and raw_data/GPS/New Machine2015-2018.xls files.

[somphavanh]

we would like to know are there any different between the app and the GPS machine , so we collected both

[choisy]

OK, thanks. Is it the same thing for 8101 that is both on the IPL server and the GPS machine?

[somphavanh]

ID 8101 , our staff didn't know the patient already sent the GPS to our server , so they went to map again

[choisy]

OK, thanks a lot.

[choisy]

Out of the 20 locations that were geolocated both on-site with the GPS devices and the with the smartphone app, 12 are less than 300 m appart, and the rest are 300m, 700m, 6km, 11km, 16km, 21km, 50km and 51 km appart. Do we have an idea of why some are so far appart??? In particular, do you think it's due to the device itself or to the person that self-report? Do you thus think we can trust the data from the smartphones?

[choisy]

For 8101, the coordinates reported by the server are 18 km appart from the coordinates reported on-site with a GPS device!!! Don't we have a problem here? Has the server method been tested?

[somphavanh]

for these 20 locations , maybe our team didn't map in the exact house or maybe the patients didn't send in their house. so i will verify again

[choisy]

As discussed, we need to continue the tests on the IPL server. As agreed,

• each time a location is reported through the IPL server, it should be done at least twice, if possible on 2 different days. This way, we'll be able to asses the precision of the system.
• that would be interesting to go and get the GPS location with the GPS device of the 11 houses that have been reported by the IPL server but not checked directly with the GPS device. This way, we'll have 31 points to compare instead of only 20.
• that would be interesting to compare the GPS locations we get through the IPL server reported by patients and reported by municipality volunteers (also through the IPL server). This way, we'll be able to assess whether the patients do something wrong when they report their locations.

Somphavanh: can you take care of this?
Thanks, Marc.

[somphavanh]

yes of courese

[choisy]

Sorry, I think my previous comment is a bit confusing. The thing is that the quality of WhatsApp data have been check on 20 points and it shows that the quality it's not that great. It would be interesting to test a few more points (10 or 20 more) to have a better idea. Also, it might be interesting to investigate why sometimes the distances are very big:

• people have several houses?
• there is a delay due to the internet connection?
• people were not at home?

These hypotheses can be tested. Then, it would be great to test the 31 points that have been collected through the server and that haven't been double-checked with the GPS device yet (30 of them).