FAQFAQ   SearchSearch  MemberlistMemberlistRegisterRegister  ProfileProfile   Log in[ Log in ]  Flint Talk RSSFlint Talk RSS

Home Open Chat Political Talk  Flint Journal Political Jokes The Bob Leonard Show  

Flint Michigan online news magazine. We have lively web forums


FlintTalk.com Forum Index > Political Talk

Topic: Lead data revisited

  Author    Post Post new topic Reply to topic
BillPayer
F L I N T O I D

With a draw down of state services coming near I've decided to look at the public lead sample data again. I 'sanitized' the data and wrote a little app to facilitate this, but so far I've only looked at a few things:

# Samples Collected: 29,067
# Unique addresses: 19,412

# of unique sites with a sample above:
5 PPB: 3518 (18.1%)
10 PPB: 2165 (11.2%)
15 PPB: 1651 (8.5%)

* It's been said that there are 30,000 water customers. With only 19,412 unique addresses that would imply 65% of customers have tested their water. HOWEVER: many apartment complexes have tested ALL of their units. These count as unique addresses, however for the 30k customer stats I believe they should only count as 1. So this 65% statistic could be very misleading.

* Only 819 (49.6%) of the unique addresses with a sample above 15 PPB tested their water more than once.

More to come. If you are interested in anything specific feel free to ask.
Post Fri Jun 30, 2017 2:32 pm 
 View user's profile Send private message  Reply with quote  
BillPayer
F L I N T O I D

Monthly analysis from public data. Total number of samples for each month as well as percentages below 5ppb and percentages greater than or equal to 5, 10, and 15 ppb.

code:
period    #samples  <  5ppb   5+ ppb    10+ ppb   15+ ppb
---------------------------------------------------------
9/2015    40        62.50%    37.50%    20.00%    12.50%
10/2015   330       73.64%    26.36%    11.21%    7.88%
11/2015   126       81.75%    18.25%    7.14%     7.14%
12/2015   176       75.57%    24.43%    13.07%    7.39%
1/2016    4980      83.07%    16.93%    9.34%     6.65%
2/2016    7900      80.46%    19.54%    11.04%    8.16%
3/2016    4916      76.87%    23.13%    13.67%    9.64%
4/2016    2478      77.44%    22.56%    12.39%    9.32%
5/2016    1421      77.41%    22.59%    12.32%    8.87%
6/2016    804       72.76%    27.24%    16.17%    12.06%
7/2016    686       75.51%    24.49%    13.85%    9.48%
8/2016    589       78.10%    21.90%    13.75%    9.34%
9/2016    609       76.68%    23.32%    11.49%    7.88%
10/2016   336       84.82%    15.18%    8.93%     5.65%
11/2016   589       86.93%    13.07%    5.26%     3.57%
12/2016   296       88.51%    11.49%    6.76%     5.41%
1/2017    289       88.58%    11.42%    5.19%     3.81%
2/2017    618       87.22%    12.78%    6.63%     4.53%
3/2017    668       88.47%    11.53%    6.59%     3.89%
4/2017    457       88.84%    11.16%    6.56%     5.03%
5/2017    462       91.77%    8.23%     4.98%     3.68%
6/2017    297       90.91%    9.09%     5.05%     3.37%


Post Mon Jul 03, 2017 8:17 am 
 View user's profile Send private message  Reply with quote  
BillPayer
F L I N T O I D

There is a large anomaly in the public residential lead test result data. A very large number of zip codes associated with results are incorrect. The number of incorrect zip codes is over 20%, and this is not a simple data entry error. That's not to say it's intentional, but this data has not been corrected in over a year.

Examples found in the data are:

For examples:
LG60791,7/20/2016,Lead,,,,0,Copper,,,,0,1606,PARK ST,FLINT,48506
LG67868,8/18/2016,Lead,,,,0,Copper,,,,0,1606,PARK ST,FLINT,48503
LG75353,9/21/2016,Lead,,,,0,Copper,,,,0,1606,PARK ST,FLINT,48504

LG60779,7/20/2016,Lead,,,,1,Copper,,,,0,1206,BLANCHARD AVE,FLINT,48529
LG87429,11/17/2016,Lead,,,,3,Copper,,,,60,1206,BLANCHARD AVE,FLINT,48507
LG75383,9/21/2016,Lead,,,,7,Copper,,,,70,1206,BLANCHARD AVE,FLINT,48507
LG24217,3/16/2016,Lead,,,,31,Copper,,,,70,1206,BLANCHARD AVE,FLINT,48503
LG29818,3/30/2016,Lead,,,,12,Copper,,,,60,1206,BLANCHARD AVE,FLINT,48503

Since there are such a large number of errors in the zip code portion, how trust worthy is the rest of the information?
Post Mon Jul 03, 2017 6:45 pm 
 View user's profile Send private message  Reply with quote  
BillPayer
F L I N T O I D

What do the zip code errors do to the data?

48506: Percentage of samples at or above the action limit was 2.53% higher than state data reflected.

48507: Percentage of samples at or above the action limit was 2.92% lower than states data reflects.



The following table has a row for each target zip code. The columns are broken up into two sets of data. 2017 samples with corrected zip codes, and 2017 data with the raw data from the states spreadsheets. Each section contains the number of samples in that zip code and the percentage at or above the action limit.

The last column indicates the change to the percentage of samples at or above the action limit. A positive percentage indicates the incorrect zip codes ADDED to the number of samples at or above the limit. A negative value indicates the zip code errors REMOVED samples at or above the action limit.

code:
          [=Corrected=====]      [=Raw Data=======]
Zip Code  # Samples   % AL+      # Samples   % AL+   Diff.
48502:      154       9.09%  =>   175        9.14%   +0.05%
48503:     1573       6.04%  =>  1600        5.44%   -0.6%
48504:     1028       7.20%  =>  1077        5.57%   -1.63%
48505:      588       6.46%  =>   588        7.99%   +1.53%
48506:      471       7.43%  =>   490        4.90%   -2.53%
48507:      741       5.53%  =>   769        8.45%   +2.92%




The samples used in this are a subset of the full 2017 tests. They are only the samples in the 2017 data which had a tests on their block before june 30th 2016. This is because the pre-june 30th data has correct zip codes. I cross referenced street names and block numbers with this data to verify correct zip codes in the 2017 data. Only samples which could be verified/corrected were used.
Post Wed Jul 05, 2017 2:04 am 
 View user's profile Send private message  Reply with quote  
  Display posts from previous:      
Post new topic Reply to topic

Jump to:  


Last Topic | Next Topic  >

Forum Rules:
You cannot post new topics in this forum
You cannot reply to topics in this forum
You cannot edit your posts in this forum
You cannot delete your posts in this forum
You cannot vote in polls in this forum

 

Flint Michigan online news magazine. We have lively web forums

Website Copyright 2010 Flint Talk.com
Contact Webmaster - FlintTalk.com >