Some addresses are very slow to Geocode

[mikepease]

Most of the time, my 2Ghz 3.5GB RAM Windows Vista laptop with Postgres 8.4.8/ PostGIS 1.5 will do a reasonably quick geocode taking from about 0.1 seconds to about 6 seconds.

But, a small handful of addresses take a very long time. About 2 to 4 minutes.

There appears to be a pattern of incorrectly spelled street names correlating to these long searches.

Here's some scenarios that appear in my data:

1) Incorrect ordinal suffix used on street number (e.g. TH instead of RD)

2) Street numbers with a ½ (e.g. 33 ½ St)

3) Pre-direction with a period (e.g. W. instead of W)

Here's a listing of specific addresses that took a long time to run:

Seconds Rating Original_Address Geocoded_Address

151.856 10 "3937 43TH AVE S, MINNEAPOLIS, MN 55406" "3937 40th Ave S, Minneapolis, MN 55406"

147.047 10 "6285 143TH LANE NW, RAMSEY, MN 55303-4635" "6285 145th Ln NW, Ramsey, MN 55303"

152.608 10 "6015 2 ½ ST NE, FRIDLEY, MN 55432" "6015 2 ½ St NE, Fridley, MN 55432"

146.024 10 "3504 33TH AVE S, MINNEAPOLIS, MN 55406" "3504 34th Ave S, Minneapolis, MN 55406"

146.475 10 "3317 32TH AVE S, MINNEAPOLIS, MN 55406" "3317 39th Ave S, Minneapolis, MN 55406"

151.847 10 "3937 43TH AVE S, MINNEAPOLIS, MN 55406" "3937 40th Ave S, Minneapolis, MN 55406"

141.642 10 "6020 6TTH ST NE, FRIDLEY, MN 55432" "6020 6th St NE, Fridley, MN 55432"

134.803 10 "2441 131TH AVE NW, COON RAPIDS, MN 55448-2554" "2441 130th Ave NW, Coon Rapids, MN 55448"

130.096 10 "200 W. 46TH ST, MINNEAPOLIS, MN 55400" "200 W 46th St, Minneapolis, MN 55419"

150.210 10 "6015 2 ½ ST NE, FRIDLEY, MN 55432" "6015 2 ½ St NE, Fridley, MN 55432"

134.722 10 "2441 131TH AVE NW, COON RAPIDS, MN 55448-2554" "2441 130th Ave NW, Coon Rapids, MN 55448"

128.609 10 "105 W. 35TH ST, MINNEAPOLIS, MN 55408" "105 W 35th St, Minneapolis, MN 55408"

129.624 10 "105 W. 35TH ST, MINNEAPOLIS, MN 55408" "105 W 35th St, Minneapolis, MN 55408"

129.164 10 "105 W. 35TH ST, MINNEAPOLIS, MN 55408" "105 W 35th St, Minneapolis, MN 55408"

136.992 10 "11 621/2 WAY, FRIDLEY, MN 55432" "11 62 ½ Way NE, Fridley, MN 55432"

138.750 12 "2011 122TH ST E, BURNSVILLE, MN 55337-4177" "2011 E 125th St, Burnsville, MN 55337"

128.056 12 "171 63TH WAY NE, FRIDLEY, MN 55432" "64th Way NE, Fridley, MN 55432"

147.613 15 "12775 73TH PL N, MAPLE GROVE, MN 55369" "f"

139.056 23 "322 W 59 ½ St, Minneapolis, MN 55419" "f"

153.953 24 "5900 25 ½ ST W, ST. LOUIS PARK, MN 55416" "f"

142.701 26 "3808 26 ½ AVE N, ROBBINSDALE, MN 55422" "f"

148.425 29 "7008 74 ½ AVE N, BROOKLYN PARK, MN 55428" "f"

146.929 29 "6601 75 ½ AVE N, BROOKLYN PARK, MN 55428" "f"

122.078 30 "501 ½ 6TH ST, NEWPORT, MN 55055" "501 12th St, Newport, MN 55055"

174.816 31 "7017-21 74 ½ AVE N, BROOKLYN PARK, MN 55428" "f"

174.876 31 "7016-20 74 ½ AVE N, BROOKLYN PARK, MN 55428" "f"

129.264 33 "317 ½ 8TH ST SE, MINNEAPOLIS, MN 55414" "f"

137.373 35 "7353 ½ 15th St N, Oakdale, MN 55128" "f"

142.824 40 "2515 ½ 16TH AVE S, MINNEAPOLIS, MN 55404" "f"

147.729 40 "3026 ½ 36TH AVE S, MINNEAPOLIS, MN 55406" "3026 36th Ave S, Minneapolis, MN 55406"

147.313 40 "3026 ½ 36TH AVE S, MINNEAPOLIS, MN 55406" "3026 36th Ave S, Minneapolis, MN 55406"

131.022 40 "142 ½ 27TH AVE SE, MINNEAPOLIS, MN 55414-3436" "142 27th Ave SE, Minneapolis, MN 55414"

[robe]

The W. issue was due to trailing spaces left in normalize address, fixed at r7507.

Next on my list is to deal with numeric street names which I have almost done but need to do some speed tests to make sure it doesn't slow anything else down.

[robe]

Mike,

I put in logic to handle mangled numeric streets at r7509 so that should take care of a good chunk in your above list (so they return the right answers and much faster).

The street num ranges and fractional street numbers is a separate thing on my list to do. We have a lot of street number ranges in our Massachusetts data so that one is kind of high, but would probably require additional field or 2 in norm_addy, which I've been putting off until I'm sure that's the best way to go.

[robe]

I think this is done. The fractional is already a separate task item in #886