| Index | Recent Threads | Unanswered Threads | Who's Active | Guidelines | Search |
 |
World Community Grid Forums
Category: Completed Research
Forum: Smash Childhood Cancer
Thread: Suddenly 91% complete [Resolved] |
| No member browsing this thread |
|
Thread Status: Active Total posts in this thread: 94
|
|
| Author |
|
|
hchc
Veteran Cruncher USA Joined: Aug 15, 2006 Post Count: 802 Status: Offline Project Badges: 
|
@knreed, thanks for the reply, but I still have a question.
----------------------------------------Your post explains how the Estimated Completion Date is calculated and why it can move up much faster in these "surge" periods. But what I'm still confused with is how the Percentage Complete is calculated and why it can suddenly jump higher or lower. That part doesn't make sense to me. Percentage Complete should simply be [amount of work completed] / [amount of total work] * 100% so it shouldn't ever go down, only up.
[Edit 1 times, last edit by hchc at Sep 16, 2018 7:32:58 AM] |
||
|
|
TonyEllis
Senior Cruncher Australia Joined: Jul 9, 2008 Post Count: 261 Status: Offline Project Badges: 
|
[amount of work completed] / [amount of total work] * 100% That only works reasonably well if the WUs are completed at a constant rate during the whole project... Suspect that WCG is trying to calculate the end date using the completion rate based on number of WUs returned over just the last few weeks; believing that to be more realistic and accurate, than basing end date on the average rate over the entire project... i.e. completion by time - not the number of WUs? Let's hope we get some clarification... Look at HSTB, for example, of how the number of WUs completed over a given period can vary massively...
Run Time Stats https://grassmere-productions.no-ip.biz/
----------------------------------------[Edit 2 times, last edit by TonyEllis at Sep 16, 2018 12:11:22 PM] |
||
|
|
hchc
Veteran Cruncher USA Joined: Aug 15, 2006 Post Count: 802 Status: Offline Project Badges:
|
@TonyEllis, your post didn't address my post. I wasn't discussing how the End Date calculation is calculated. My question was why the Percent Complete sometimes goes down, when it shouldn't ever go down.
----------------------------------------
|
||
|
|
Sgt.Joe
Ace Cruncher USA Joined: Jul 4, 2006 Post Count: 7670 Status: Offline Project Badges: 
|
@TonyEllis, your post didn't address my post. I wasn't discussing how the End Date calculation is calculated. My question was why the Percent Complete sometimes goes down, when it shouldn't ever go down. The most obvious reason the percentage may go down is the scientists may have more work than they initially have anticipated. If their initial estimate for their work is 100,000,000 work units and then they decide they have the (money, time, resources, new insights, new libraries to test against, etc.) and then include an additional 100,000,000 work units, then the percentage will decrease. Cheers
Sgt. Joe
*Minnesota Crunchers* |
||
|
|
hchc
Veteran Cruncher USA Joined: Aug 15, 2006 Post Count: 802 Status: Offline Project Badges:
|
Sgt Joe said:
----------------------------------------@TonyEllis, your post didn't address my post. I wasn't discussing how the End Date calculation is calculated. My question was why the Percent Complete sometimes goes down, when it shouldn't ever go down. The most obvious reason the percentage may go down is the scientists may have more work than they initially have anticipated. If their initial estimate for their work is 100,000,000 work units and then they decide they have the (money, time, resources, new insights, new libraries to test against, etc.) and then include an additional 100,000,000 work units, then the percentage will decrease. Cheers Yep, I get that. So is it safe to say that the 91% completed number is accurate? I'm trying to decide if I should scramble and allocate all resources to SCC if we're that close to 100% completion.
|
||
|
|
TonyEllis
Senior Cruncher Australia Joined: Jul 9, 2008 Post Count: 261 Status: Offline Project Badges:
|
Sorry if I didn't make it clear enough... I was hinting the end-date and completion percentage could be by time completed - not number of WUs... let's try a simple example... calculate percent completion every 32 days... time using the rate WUs returned over the previous 32 days...
----------------------------------------We have a project with 256 WUs and complete a constant 1 WU per day for the first 128 WUS. So at the end of 128 days - we have completed 50% by time and by # WUs. Then over the next 32 days we only complete at constant rate one WUs every two days i.e. 16 WUs.. So at the end of 160 (128+32) days we have completed 144 (128+16) WUS... i.e. (144/256) x 100 = 56.25% by # WUs. So the completion by # WUs has increased. However, If we were to continue at the rate 1 WU per 2 days for the rest of the WUS we would take (256 - 144) x 2 = 224 days more to complete. Doing the calculation for completion by time (160/(224 +160)) x 100 = 42.78 % complete, where as at 128 days we were 50% complete by time.. so the completion percentage went down  The end date would be more into the future than at 128 days into the project... and as indicated - the number of WUs for the total project could increase or decrease resulting in a sudden change in completion percentage...
Run Time Stats https://grassmere-productions.no-ip.biz/
|
||
|
|
hchc
Veteran Cruncher USA Joined: Aug 15, 2006 Post Count: 802 Status: Offline Project Badges:
|
TonyEllis said:
----------------------------------------Sorry if I didn't make it clear enough... I was hinting the end-date and completion percentage could be by time completed - not number of WUs... let's try a simple example... calculate percent completion every 32 days... time using the rate WUs returned over the previous 32 days... We have a project with 256 WUs and complete a constant 1 WU per day for the first 128 WUS. So at the end of 128 days - we have completed 50% by time and by # WUs. Then over the next 32 days we only complete at constant rate one WUs every two days i.e. 16 WUs.. So at the end of 160 (128+32) days we have completed 144 (128+16) WUS... i.e. (144/256) x 100 = 56.25% by # WUs. 144 work units completed / 256 total work units * 100% = 56.25% Full stop. TonyEllis said: So the completion by # WUs has increased. However, If we were to continue at the rate 1 WU per 2 days for the rest of the WUS we would take (256 - 144) x 2 = 224 days more to complete. Doing the calculation for completion by time (160/(224 +160)) x 100 = 42.78 % complete, where as at 128 days we were 50% complete by time.. so the completion percentage went down The end date would be more into the future than at 128 days into the project... and as indicated - the number of WUs for the total project could increase or decrease resulting in a sudden change in completion percentage... The above paragraph ("completion by time") is supremely confusing and doesn't communicate what most people think of when talking about percentages. Percentage is simply a raw calculation...i.e., "How much is done?" and shouldn't introduce variables such as time. Time only matters when calculating ETA. Time should not matter when calculating Percentage completion...that should simply be a raw calculation of [Work Units completed] / [Total known Work Units] * 100%. By that math, the percentage should only go down if more batches are introduced into the project because of new funding, etc.
|
||
|
|
Speedy51
Veteran Cruncher New Zealand Joined: Nov 4, 2005 Post Count: 1293 Status: Offline Project Badges: 
|
hchc please refer to Knreed post in https://www.worldcommunitygrid.org/forums/wcg...ad,40597_offset,20#593628 for the answer to why the progress has jumped so much on the research page
---------------------------------------- |
||
|
|
hchc
Veteran Cruncher USA Joined: Aug 15, 2006 Post Count: 802 Status: Offline Project Badges:
|
Speedy51 said:
----------------------------------------hchc please refer to Knreed post in https://www.worldcommunitygrid.org/forums/wcg...ad,40597_offset,20#593628 for the answer to why the progress has jumped so much on the research page I don't believe you're following the conversation. The thread you linked simply linked back to this thread, specifically knreed's explanation. And if you read this thread, I replied to knreed's post where I asked: hchc said: @knreed, thanks for the reply, but I still have a question. Your post explains how the Estimated Completion Date is calculated and why it can move up much faster in these "surge" periods. But what I'm still confused with is how the Percentage Complete is calculated and why it can suddenly jump higher or lower. That part doesn't make sense to me. Percentage Complete should simply be [amount of work completed] / [amount of total work] * 100% so it shouldn't ever go down, only up. So Speedy51, I've read the explanations, and I had follow-up questions specifically on how Percentage complete is calculated, not how ETA is calculated. My confusion isn't related to ETA, it's simply asking if the 91% complete number is in fact accurate, one shouldn't expect it to decrease. In other words, 91% is accurate.
[Edit 1 times, last edit by hchc at Sep 17, 2018 3:39:37 AM] |
||
|
|
TonyEllis
Senior Cruncher Australia Joined: Jul 9, 2008 Post Count: 261 Status: Offline Project Badges:
|
Time should not matter when calculating Percentage completion. We will have to agree to disagree - and this is my last post... When building a house - when would most people consider it 50% complete i.e.half way there? by *time* - or the number of items used in its construction, man hours worked or...
Run Time Stats https://grassmere-productions.no-ip.biz/
|
||
|
|
|